| summaryrefslogtreecommitdiff |
diff options
| author | Nathanael Sensfelder <SpamShield0@MultiAgentSystems.org> | 2017-01-18 19:09:16 +0100 |
|---|---|---|
| committer | Nathanael Sensfelder <SpamShield0@MultiAgentSystems.org> | 2017-01-18 19:09:16 +0100 |
| commit | 0d49fb74eadcf933f696420cd182077927680d26 (patch) | |
| tree | 9220d260ce878f369138da12dae0300cf9ade5c9 | |
| parent | 24afb3e60bafd98e6a83dcb41ee6a7f7d41e76bc (diff) | |
Done with 'core', starting to work on 'knowledge'.
| -rw-r--r-- | src/CMakeLists.txt | 10 | ||||
| -rw-r--r-- | src/cli/CMakeLists.txt | 6 | ||||
| -rw-r--r-- | src/cli/cli.h (renamed from src/io/error.h) | 13 | ||||
| -rw-r--r-- | src/cli/parameters.c (renamed from src/io/parameters.c) | 0 | ||||
| -rw-r--r-- | src/cli/parameters.h (renamed from src/io/parameters.h) | 0 | ||||
| -rw-r--r-- | src/cli/parameters_types.h | 67 | ||||
| -rw-r--r-- | src/core/CMakeLists.txt | 9 | ||||
| -rw-r--r-- | src/core/assimilate.c | 281 | ||||
| -rw-r--r-- | src/core/char.c | 18 | ||||
| -rw-r--r-- | src/core/char.h | 35 | ||||
| -rw-r--r-- | src/core/char_types.h | 6 | ||||
| -rw-r--r-- | src/core/index.c | 61 | ||||
| -rw-r--r-- | src/core/index.h | 11 | ||||
| -rw-r--r-- | src/core/index_types.h | 2 | ||||
| -rw-r--r-- | src/core/sequence.c | 84 | ||||
| -rw-r--r-- | src/core/sequence.h | 20 | ||||
| -rw-r--r-- | src/core/sequence_creation.c | 9 | ||||
| -rw-r--r-- | src/core/sequence_from_string.c | 315 | ||||
| -rw-r--r-- | src/core/sequence_types.h | 3 | ||||
| -rw-r--r-- | src/file/data_input.c (renamed from src/io/data_input.c) | 0 | ||||
| -rw-r--r-- | src/file/data_input.h (renamed from src/io/data_input.h) | 0 | ||||
| -rw-r--r-- | src/file/data_input_types.h (renamed from src/io/data_input_types.h) | 0 | ||||
| -rw-r--r-- | src/file/data_output.c (renamed from src/io/data_output.c) | 1 | ||||
| -rw-r--r-- | src/file/data_output.h (renamed from src/io/data_output.h) | 0 | ||||
| -rw-r--r-- | src/io/CMakeLists.txt | 9 | ||||
| -rw-r--r-- | src/io/parameters_types.h | 21 | ||||
| -rw-r--r-- | src/irc/network.c (renamed from src/io/network.c) | 0 | ||||
| -rw-r--r-- | src/irc/network.h (renamed from src/io/network.h) | 0 | ||||
| -rw-r--r-- | src/irc/network_types.h (renamed from src/io/network_types.h) | 0 | ||||
| -rw-r--r-- | src/knowledge/CMakeLists.txt | 11 | ||||
| -rw-r--r-- | src/knowledge/knowledge.c (renamed from src/core/knowledge.c) | 8 | ||||
| -rw-r--r-- | src/knowledge/knowledge.h (renamed from src/core/knowledge.h) | 27 | ||||
| -rw-r--r-- | src/knowledge/knowledge_finalize.c (renamed from src/core/knowledge_finalize.c) | 7 | ||||
| -rw-r--r-- | src/knowledge/knowledge_learn_sequence.c | 324 | ||||
| -rw-r--r-- | src/knowledge/knowledge_learn_word.c | 276 | ||||
| -rw-r--r-- | src/knowledge/knowledge_search.c (renamed from src/core/knowledge_search.c) | 7 | ||||
| -rw-r--r-- | src/knowledge/knowledge_types.h (renamed from src/core/knowledge_types.h) | 5 | ||||
| -rw-r--r-- | src/main.c (renamed from src/core/main.c) | 0 | ||||
| -rw-r--r-- | src/pervasive.h | 34 |
39 files changed, 1267 insertions, 413 deletions
diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt index 76a73ed..594b14f 100644 --- a/src/CMakeLists.txt +++ b/src/CMakeLists.txt @@ -1,5 +1,13 @@ +add_subdirectory(cli) add_subdirectory(core) -add_subdirectory(io) +add_subdirectory(file) +add_subdirectory(irc) +add_subdirectory(knowledge) add_subdirectory(tool) +set( + SRC_FILES ${SRC_FILES} + ${CMAKE_CURRENT_SOURCE_DIR}/main.c +) + set(SRC_FILES ${SRC_FILES} PARENT_SCOPE) diff --git a/src/cli/CMakeLists.txt b/src/cli/CMakeLists.txt new file mode 100644 index 0000000..94e6337 --- /dev/null +++ b/src/cli/CMakeLists.txt @@ -0,0 +1,6 @@ +set( + SRC_FILES ${SRC_FILES} + ${CMAKE_CURRENT_SOURCE_DIR}/parameters.c +) +set(SRC_FILES ${SRC_FILES} PARENT_SCOPE) + diff --git a/src/io/error.h b/src/cli/cli.h index be7359f..5aec25b 100644 --- a/src/io/error.h +++ b/src/cli/cli.h @@ -1,15 +1,10 @@ -#ifndef _ZoO_IO_ERROR_H_ -#define _ZoO_IO_ERROR_H_ +#ifndef _ZoO_CLI_CLI_H_ +#define _ZoO_CLI_CLI_H_ #include <stdio.h> #include "../pervasive.h" -#define ZoO_DEBUG_ALL 1 - -#ifndef ZoO_DEBUG_ALL - #define ZoO_DEBUG_ALL 0 -#endif #ifndef ZoO_DEBUG_PROGRAM_FLOW #define ZoO_DEBUG_PROGRAM_FLOW (0 || ZoO_DEBUG_ALL) @@ -23,7 +18,9 @@ #define ZoO_DEBUG_LEARNING (0 || ZoO_DEBUG_ALL) #endif -#define ZoO_DEBUG_NETWORK 1 +#ifndef ZoO_DEBUG_NETWORK + #define ZoO_DEBUG_NETWORK 1 +#endif #ifndef ZoO_DEBUG_NETWORK #define ZoO_DEBUG_NETWORK (0 || ZoO_DEBUG_ALL) diff --git a/src/io/parameters.c b/src/cli/parameters.c index 77c33aa..77c33aa 100644 --- a/src/io/parameters.c +++ b/src/cli/parameters.c diff --git a/src/io/parameters.h b/src/cli/parameters.h index 1011e2b..1011e2b 100644 --- a/src/io/parameters.h +++ b/src/cli/parameters.h diff --git a/src/cli/parameters_types.h b/src/cli/parameters_types.h new file mode 100644 index 0000000..15b5254 --- /dev/null +++ b/src/cli/parameters_types.h @@ -0,0 +1,67 @@ +#ifndef _ZoO_IO_PARAMETERS_TYPES_H_ +#define _ZoO_IO_PARAMETERS_TYPES_H_ + +#include "../pervasive.h" + +/******************************************************************************/ +/** DEFAULT VALUES ************************************************************/ +/******************************************************************************/ + +#ifndef ZoO_DEFAULT_DATA_FILENAME + #define ZoO_DEFAULT_DATA_FILENAME "./memory.txt" +#endif + +#ifndef ZoO_DEFAULT_IRC_SERVER_ADDR + #define ZoO_DEFAULT_IRC_SERVER_ADDR "irc.foonetic.net" +#endif + +#ifndef ZoO_DEFAULT_IRC_SERVER_PORT + #define ZoO_DEFAULT_IRC_SERVER_PORT "6667" +#endif + +#ifndef ZoO_DEFAULT_IRC_SERVER_CHANNEL + #define ZoO_DEFAULT_IRC_SERVER_CHANNEL "#theborghivemind" +#endif + +#ifndef ZoO_DEFAULT_IRC_USERNAME + #define ZoO_DEFAULT_IRC_USERNAME "zeroofone" +#endif + +#ifndef ZoO_DEFAULT_IRC_REALNAME + #define ZoO_DEFAULT_IRC_REALNAME "Zero of One (bot)" +#endif + +#ifndef ZoO_DEFAULT_REPLY_RATE + #define ZoO_DEFAULT_REPLY_RATE 8 +#endif + +/******************************************************************************/ +/** DEBUG LEVELS **************************************************************/ +/******************************************************************************/ + +#ifndef ZoO_DEBUG_PARAMETERS + #define ZoO_DEBUG_PARAMETERS (0 || ZoO_DEBUG_ALL) +#endif + +/******************************************************************************/ +/** FUNCTIONS *****************************************************************/ +/******************************************************************************/ + +struct ZoO_parameters +{ + const char * restrict data_filename; + const char * restrict new_data_filename; + + const char * restrict irc_server_addr; + const char * restrict irc_server_port; + const char * restrict irc_server_channel; + const char * restrict irc_username; + const char * restrict irc_realname; + + int reply_rate; + + int aliases_count; + const char * restrict * restrict aliases; +}; + +#endif diff --git a/src/core/CMakeLists.txt b/src/core/CMakeLists.txt index fe28080..1e1daa8 100644 --- a/src/core/CMakeLists.txt +++ b/src/core/CMakeLists.txt @@ -1,13 +1,10 @@ set( SRC_FILES ${SRC_FILES} ${CMAKE_CURRENT_SOURCE_DIR}/char.c - ${CMAKE_CURRENT_SOURCE_DIR}/main.c - ${CMAKE_CURRENT_SOURCE_DIR}/knowledge.c - ${CMAKE_CURRENT_SOURCE_DIR}/knowledge_search.c - ${CMAKE_CURRENT_SOURCE_DIR}/knowledge_finalize.c - ${CMAKE_CURRENT_SOURCE_DIR}/assimilate.c - ${CMAKE_CURRENT_SOURCE_DIR}/sequence_creation.c + ${CMAKE_CURRENT_SOURCE_DIR}/index.c ${CMAKE_CURRENT_SOURCE_DIR}/sequence.c + ${CMAKE_CURRENT_SOURCE_DIR}/sequence_creation.c + ${CMAKE_CURRENT_SOURCE_DIR}/sequence_from_string.c ) set(SRC_FILES ${SRC_FILES} PARENT_SCOPE) diff --git a/src/core/assimilate.c b/src/core/assimilate.c deleted file mode 100644 index 7f03e1b..0000000 --- a/src/core/assimilate.c +++ /dev/null @@ -1,281 +0,0 @@ -#include <stdlib.h> -#include <string.h> - -#include "../io/error.h" - -#include "knowledge.h" - -/** Functions to assimilate sentences using a ZoO_knowledge structure *********/ - - -static int add_sequence -( - ZoO_index links_count [const], - struct ZoO_knowledge_link * links [const], - ZoO_index const sequence [const restrict static ZoO_MARKOV_ORDER], - ZoO_index const target_i, - ZoO_index const offset -) -{ - ZoO_index link_index, i; - struct ZoO_knowledge_link * link; - ZoO_index * new_p; - - if - ( - ZoO_knowledge_get_link - ( - links_count, - links, - (sequence + offset), - &link_index - ) < 0 - ) - { - return -1; - } - - link = (*links + link_index); - link->occurrences += 1; - - for (i = 0; i < link->targets_count; ++i) - { - if (link->targets[i] == sequence[target_i]) - { - link->targets_occurrences[i] += 1; - - return 0; - } - } - - link->targets_count += 1; - - new_p = - (ZoO_index *) realloc - ( - (void *) link->targets, - (sizeof(ZoO_index) * link->targets_count) - ); - - if (new_p == (ZoO_index *) NULL) - { - link->targets_count -= 1; - - /* TODO: err. */ - return -1; - } - - link->targets = new_p; - link->targets[link->targets_count - 1] = sequence[target_i]; - - new_p = - (ZoO_index *) realloc - ( - (void *) link->targets_occurrences, - (sizeof(ZoO_index) * link->targets_count) - ); - - if (new_p == (ZoO_index *) NULL) - { - link->targets_count -= 1; - - /* TODO: err. */ - return -1; - } - - link->targets_occurrences = new_p; - link->targets_occurrences[link->targets_count - 1] = 1; - - return 0; -} - -static int add_word_occurrence -( - struct ZoO_knowledge k [const restrict static 1], - ZoO_index const sequence [const static ((ZoO_MARKOV_ORDER * 2) + 1)] -) -{ - ZoO_index w; - int error; - - w = sequence[ZoO_MARKOV_ORDER]; - - error = - add_sequence - ( - &(k->words[w].forward_links_count), - &(k->words[w].forward_links), - sequence + (ZoO_MARKOV_ORDER + 1), - (ZoO_MARKOV_ORDER - 1), - 0 - ); - - error = - ( - add_sequence - ( - &(k->words[w].backward_links_count), - &(k->words[w].backward_links), - sequence, - 0, - 1 - ) - | error - ); - - return error; -} - -static int should_assimilate -( - struct ZoO_strings string [const restrict static 1], - ZoO_index const aliases_count, - const char * restrict aliases [const restrict static aliases_count] -) -{ - ZoO_index i; - - /* Don't assimilate empty strings. */ - if (string->words_count == 0) - { - return 0; - } - - /* Don't assimilate things that start with our name. */ - for (i = 0; i < aliases_count; ++i) - { - if (ZoO_IS_PREFIX(aliases[i], string->words[0])) - { - return 0; - } - } - - return 1; -} - -static int init_sequence -( - struct ZoO_knowledge k [const static 1], - struct ZoO_strings string [const restrict static 1], - ZoO_index sequence [const restrict static ((ZoO_MARKOV_ORDER * 2) + 1)] -) -{ - ZoO_index i; - - /* We are going to link this sequence to ZoO_WORD_START_OF_LINE */ - sequence[ZoO_MARKOV_ORDER] = ZoO_WORD_START_OF_LINE; - - for (i = 1; i <= ZoO_MARKOV_ORDER; ++i) - { - sequence[ZoO_MARKOV_ORDER - i] = ZoO_WORD_START_OF_LINE; - - if (i <= string->words_count) - { - if - ( - ZoO_knowledge_learn - ( - k, - string->words[i - 1], - (sequence + (ZoO_MARKOV_ORDER + i)) - ) < 0 - ) - { - return -1; - } - } - else - { - sequence[ZoO_MARKOV_ORDER + i] = ZoO_WORD_END_OF_LINE; - } - } - - return 0; -} - -int ZoO_knowledge_assimilate -( - struct ZoO_knowledge k [const static 1], - struct ZoO_strings string [const restrict static 1], - ZoO_index const aliases_count, - const char * restrict aliases [const restrict static aliases_count] -) -{ - int error; - ZoO_index sequence[(ZoO_MARKOV_ORDER * 2) + 1]; - ZoO_index next_word, new_word, new_word_id; - - if (!should_assimilate(string, aliases_count, aliases)) - { - return 0; - } - - if (init_sequence(k, string, sequence) < 0) - { - return -1; - } - - if (add_word_occurrence(k, sequence) < 0) - { - error = -1; - - /* There's a pun... */ - ZoO_S_WARNING("Could not add a link between words."); - - return -1; - } - - error = 0; - - next_word = 0; - new_word = ZoO_MARKOV_ORDER; - - while (next_word <= (string->words_count + ZoO_MARKOV_ORDER)) - { - if (new_word < string->words_count) - { - /* prevents words [restrict], k [restrict] */ - if (ZoO_knowledge_learn(k, string->words[new_word], &new_word_id) < 0) - { - return -1; - } - } - else - { - new_word_id = ZoO_WORD_END_OF_LINE; - } - - memmove - ( - (void *) sequence, - (const void *) (sequence + 1), - /* Accepts 0. */ - (sizeof(ZoO_index) * (ZoO_MARKOV_ORDER * 2)) - ); - - sequence[ZoO_MARKOV_ORDER * 2] = new_word_id; - - if (add_word_occurrence(k, sequence) < 0) - { - error = -1; - - /* There's a pun... */ - ZoO_S_WARNING("Could not add a link between words."); - - return -1; - } - - /* - * Safe: - * - next_word < words_count - * - words_count =< ZoO_INDEX_MAX - * ---- - * next_word < ZoO_INDEX_MAX - */ - next_word += 1; - new_word += 1; - } - - return error; -} - diff --git a/src/core/char.c b/src/core/char.c index 39ca72e..9297643 100644 --- a/src/core/char.c +++ b/src/core/char.c @@ -2,6 +2,18 @@ #include "char.h" +/* See: "char.c" */ +ZoO_char ZoO_char_to_lowercase (const ZoO_char c) +{ + if ((c >= 'A') && (c <= 'Z')) + { + return 'z' - ('Z' - c); + } + + return c; +} + +/* See: "char.c" */ int ZoO_char_is_banned (const ZoO_char c) { switch (c) @@ -21,6 +33,7 @@ int ZoO_char_is_banned (const ZoO_char c) } } +/* See: "char.c" */ int ZoO_char_is_punctuation (const ZoO_char c) { switch (c) @@ -38,11 +51,14 @@ int ZoO_char_is_punctuation (const ZoO_char c) } } +/* See: "char.c" */ int ZoO_word_cmp ( const ZoO_char word_a [const static 1], + const size_t word_a_size, const ZoO_char word_b [const static 1] ) { - return strcmp((const char *) word_a, (const char *) word_b); + return strncmp((const char *) word_a, (const char *) word_b, word_a_size); } + diff --git a/src/core/char.h b/src/core/char.h index 772a3a2..2b4a355 100644 --- a/src/core/char.h +++ b/src/core/char.h @@ -3,19 +3,42 @@ #include "char_types.h" -enum ZoO_word_property ZoO_get_word_property -( - const ZoO_char word [const restrict], - size_t word_size -); - +/* Compares two words. {word_a} does not have to be null terminated. */ +/*@ + @ requires null_terminated_string(word_b); + @ requires ((length(word_a) * sizeof(ZoO_char)) == word_a_size); + @ ensures ((\result == 1) || (\result == 0) || (\result == -1)); + @*/ int ZoO_word_cmp ( const ZoO_char word_a [const static 1], + const size_t word_a_size, const ZoO_char word_b [const static 1] ); +/* + * Returns the lowercase equivalent of ZoO_char that are included in ['A','Z']. + * Other ZoO_char are returned untouched. + */ +ZoO_char ZoO_char_to_lowercase (const ZoO_char c); + +/* + * Returns '1' iff {c} should be considered as an punctuation character, '0' + * otherwise. + */ +/*@ + @ ensures ((\result == 1) || (\result == 0)); + @*/ int ZoO_char_is_punctuation (const ZoO_char c); + +/* + * Returns '1' iff containing {c} means the word should not be learned. '0' + * otherwise. + */ +/*@ + @ ensures ((\result == 1) || (\result == 0)); + @*/ int ZoO_word_char_is_banned (const ZoO_char c); #endif + diff --git a/src/core/char_types.h b/src/core/char_types.h index 67b5294..a2a736c 100644 --- a/src/core/char_types.h +++ b/src/core/char_types.h @@ -1,12 +1,16 @@ #ifndef _ZoO_CORE_CHAR_TYPES_H_ #define _ZoO_CORE_CHAR_TYPES_H_ - +/* + * FIXME: Does not belong here. + */ +/* enum ZoO_word_property { ZoO_WORD_NO_PROPERTY, ZoO_WORD_HAS_NO_LEFT_SEPARATOR, ZoO_WORD_HAS_NO_RIGHT_SEPARATOR }; +*/ /* ZoO_char = UTF-8 char */ typedef char ZoO_char; diff --git a/src/core/index.c b/src/core/index.c new file mode 100644 index 0000000..375e0ad --- /dev/null +++ b/src/core/index.c @@ -0,0 +1,61 @@ +#include <limits.h> +#include <stdlib.h> + +#include "index.h" + +#if (RAND_MAX < UCHAR_MAX) + #error "RAND_MAX < UCHAR_MAX, unable to generate random numbers." +#endif + +#if (RAND_MAX == 0) + #error "RAND_MAX is included in [0, 0]. What are you even doing?" +#endif + +/* + * Returns a random unsigned char. + */ +static unsigned char random_uchar (void) +{ + return + (unsigned char) + ( + /* FIXME: Do floats allow enough precision for this? */ + ( + ((float) rand()) + / ((float) RAND_MAX) + ) + * ((float) UCHAR_MAX) + ); +} + +/* See: "index.h" */ +ZoO_index ZoO_index_random (void) +{ + ZoO_index i; + ZoO_index result; + unsigned char * result_bytes; + + result_bytes = (unsigned char *) &result; + + for (i = 0; i < sizeof(ZoO_index); ++i) + { + result_bytes[i] = random_uchar(); + } + + return result; +} + +/* See: "index.h" */ +ZoO_index ZoO_index_random_up_to (const ZoO_index max) +{ + return + (ZoO_index) + ( + /* FIXME: Do floats allow enough precision for this? */ + ( + ((float) ZoO_index_random()) + / ((float) ZoO_INDEX_MAX) + ) + * ((float) max) + ); +} diff --git a/src/core/index.h b/src/core/index.h index 76e3507..1417662 100644 --- a/src/core/index.h +++ b/src/core/index.h @@ -3,6 +3,17 @@ #include "index_types.h" +/* + * Returns a random ZoO_index. + */ +ZoO_index ZoO_index_random (void); + +/* + * Returns a random ZoO_index, included in [0, limit] + */ +/*@ + @ ensures (\result <= limit); + @*/ ZoO_index ZoO_index_random_up_to (const ZoO_index limit); #endif diff --git a/src/core/index_types.h b/src/core/index_types.h index 2d769ca..ad56d52 100644 --- a/src/core/index_types.h +++ b/src/core/index_types.h @@ -1,8 +1,10 @@ #ifndef _ZoO_CORE_INDEX_TYPES_H_ #define _ZoO_CORE_INDEX_TYPES_H_ +/* Must be unsigned. */ typedef unsigned int ZoO_index; +/* Must be > 0. */ #define ZoO_INDEX_MAX UINT_MAX #endif diff --git a/src/core/sequence.c b/src/core/sequence.c index 9e370a3..d7ff9d0 100644 --- a/src/core/sequence.c +++ b/src/core/sequence.c @@ -5,18 +5,56 @@ #include "sequence.h" +/* + * Bypass rendundant ZoO_START_OF_SEQUENCE_ID at the start of a sequence. + */ +/* ensures (*sequence_offset <= sequence_length) */ +static void bypass_redundant_sos +( + const ZoO_index sequence [const restrict], + const ZoO_index sequence_length, + ZoO_index sequence_offset [const restrict static 1] +) +{ + ZoO_index i; + + *sequence_offset = 0; + + for (i = 0; i < sequence_length; ++i) + { + if (sequence[i] != ZoO_START_OF_SEQUENCE_ID) + { + return; + } + else if (sequence[i] == ZoO_START_OF_SEQUENCE_ID) + { + *sequence_offset = i; + } + } +} + + /* See "sequence.h" */ int ZoO_sequence_cmp ( const ZoO_index sequence_a [const], - const ZoO_index sequence_a_length, + ZoO_index sequence_a_length, const ZoO_index sequence_b [const], - const ZoO_index sequence_b_length + ZoO_index sequence_b_length ) { - ZoO_index min_length; + ZoO_index min_length, a, b; + ZoO_index a_offset, b_offset; ZoO_index i; + bypass_redundant_sos(sequence_a, sequence_a_length, &a_offset); + bypass_redundant_sos(sequence_b, sequence_b_length, &b_offset); + + /*@ requires (*a_offset <= sequence_a_length) @*/ + sequence_a_length -= a_offset; + /*@ requires (*b_offset <= sequence_b_length) @*/ + sequence_b_length -= b_offset; + if (sequence_a_length < sequence_b_length) { min_length = sequence_a_length; @@ -26,47 +64,37 @@ int ZoO_sequence_cmp min_length = sequence_b_length; } + /*@ ensures (min_length <= sequence_a_length) @*/ + /*@ ensures (min_length <= sequence_b_length) @*/ + for (i = 0; i < min_length; ++i) { - if (sequence_a[i] < sequence_b[i]) + /*@ requires ((i + a_offset) < sequence_a_length) @*/ + a = sequence_a[i + a_offset]; + /*@ requires ((i + b_offset) < sequence_b_length) @*/ + b = sequence_b[i + b_offset]; + + if (a < b) { return -1; } - else if (sequence_b[i] > sequence_b[i]) + else if (b > a) { return 1; } - else if - ( - (sequence_a[i] == ZoO_END_OF_SEQUENCE_ID) - && (sequence_b[i] == ZoO_END_OF_SEQUENCE_ID) - ) + else if ((a == ZoO_END_OF_SEQUENCE_ID) && (b == ZoO_END_OF_SEQUENCE_ID)) { return 0; } } - if (sequence_a_length < sequence_b_length) + if (sequence_a_length > sequence_b_length) { - if (sequence_b[i] == ZoO_END_OF_SEQUENCE_ID) - { - return 0; - } - else - { - return -1; - } + return 1; } - else if (sequence_a_length > sequence_b_length) + else if (sequence_a_length < sequence_b_length) { - if (sequence_a[i] == ZoO_END_OF_SEQUENCE_ID) - { - return 0; - } - else - { - return 1; - } + return -1; } else { diff --git a/src/core/sequence.h b/src/core/sequence.h index e609b4d..77ecd6c 100644 --- a/src/core/sequence.h +++ b/src/core/sequence.h @@ -1,11 +1,21 @@ #ifndef _ZoO_CORE_SEQUENCE_H_ #define _ZoO_CORE_SEQUENCE_H_ +#include "../core/char_types.h" #include "../core/index_types.h" -#include "../core/knowledge_types.h" +#include "../knowledge/knowledge_types.h" #include "sequence_types.h" +int ZoO_sequence_from_undercase_string +( + const ZoO_char string [const restrict], + const ZoO_index string_length, + struct ZoO_knowledge k [const restrict static 1], + ZoO_index * sequence [const restrict static 1], + ZoO_index sequence_length [const restrict static 1] +); + /* * Creates a sequence containing {initial_word}. The remaining elements of * sequence are added according to what is known to {k} as being possible. @@ -42,7 +52,13 @@ int ZoO_sequence_create_from * ZoO_END_OF_SEQUENCE marks the ending of a sequence, regardless of indicated * sequence length, meaning that [10][ZoO_END_OF_SEQUENCE][9] and * [10][ZoO_END_OF_SEQUENCE][8] are considered equal. Sequences do not have to - * contain ZoO_END_OF_SEQUENCE. + * contain ZoO_END_OF_SEQUENCE. [10][ZoO_END_OF_SEQUENCE] and [10] are + * considered different, [10][ZoO_END_OF_SEQUENCE] + * and [10][ZoO_END_OF_SEQUENCE][ZoO_END_OF_SEQUENCE] are considered equal. + * Same logic is applyied for ZoO_START_OF_SEQUENCE: + * [START_OF_SEQUENCE][10] is not [10], but + * [START_OF_SEQUENCE][START_OF_SEQUENCE][10] and [START_OF_SEQUENCE][10] are + * the same. * Return: * 1 iff {sequence_a} should be considered being more than {sequence_b} * 0 iff {sequence_a} should be considered being equal to {sequence_b} diff --git a/src/core/sequence_creation.c b/src/core/sequence_creation.c index 1133be9..f460629 100644 --- a/src/core/sequence_creation.c +++ b/src/core/sequence_creation.c @@ -19,6 +19,11 @@ * (> weights_sum 0). * (= (sum weights) weights_sum). */ +/*@ + @ requires (weights_sum > 0); + @ requires \valid(weights); + @ requires (\sum(0, (\length(weights) - 1), weights) = weights_sum); +@*/ static ZoO_index weighted_random_pick ( const ZoO_index weights [const restrict static 1], @@ -29,12 +34,12 @@ static ZoO_index weighted_random_pick accumulator = 0; - /* Safe: Included in [0, weights_sum]. */ random_number = ZoO_index_random_up_to(weights_sum); + /*@ ensures (0 <= random_number <= weights_sum); @*/ for (result = 0; accumulator < random_number; ++result) { - /* Safe: (= (sum weights) weights_sum) */ + /*@ requires (\sum(0, (\length(weights) - 1), weights) = weights_sum); @*/ accumulator += weights[result]; } diff --git a/src/core/sequence_from_string.c b/src/core/sequence_from_string.c new file mode 100644 index 0000000..51d7049 --- /dev/null +++ b/src/core/sequence_from_string.c @@ -0,0 +1,315 @@ +#define _POSIX_C_SOURCE 200809L +#include <stdlib.h> +#include <string.h> +#include <stdint.h> /* defines SIZE_MAX */ + +#include "../core/char.h" +#include "../core/index.h" + +#include "../cli/cli.h" + +#include "../knowledge/knowledge.h" + +#include "sequence.h" + +static int add_word_id_to_sequence +( + const ZoO_index word_id, + ZoO_index * sequence [const restrict static 1], + ZoO_index sequence_length [const restrict static 1] +) +{ + ZoO_index * new_sequence; + + *sequence_length += 1; + + new_sequence = + (ZoO_index *) realloc + ( + (void *) *sequence, + (((size_t) sequence_length) * sizeof(ZoO_index)) + ); + + if (new_sequence == (ZoO_index *) NULL) + { + ZoO_S_ERROR("Unable to reallocate a sequence to add word ids to it."); + + return -1; + } + + return 0; +} + +/******************************************************************************/ +/** HANDLING PUNCTUATION ******************************************************/ +/******************************************************************************/ +static int add_punctuation_to_sequence +( + const ZoO_char string [const restrict static 1], + const ZoO_char punctuation, + ZoO_index * sequence [const restrict static 1], + ZoO_index sequence_length [const restrict static 1], + const struct ZoO_knowledge k [const restrict static 1] +) +{ + ZoO_index word_id; + ZoO_char as_word[2]; + + as_word[0] = punctuation; + as_word[1] = '\0'; + + if (ZoO_knowledge_find_word_id(k, as_word, 2, &word_id) < 0) + { + ZoO_PROG_ERROR + ( + "'%s' was defined as a punctuation, was found in a string, yet is not" + " defined in the knowledge database.", + as_word + ); + + return -1; + } + + if (add_word_id_to_sequence(word_id, sequence, sequence_length) < 0) + { + return -1; + } + + return 0; +} + +static int word_is_punctuation_terminated +( + const ZoO_char string [const restrict static 1], + const ZoO_index word_start, + const ZoO_index word_length +) +{ + return ZoO_char_is_punctuation(string[word_length]); +} + +/******************************************************************************/ +/** HANDLING WORDS ************************************************************/ +/******************************************************************************/ +static int add_word_to_sequence +( + const ZoO_char string [const restrict static 1], + const ZoO_index word_start, + const ZoO_index word_length, + ZoO_index * sequence [const restrict static 1], + ZoO_index sequence_length [const restrict static 1], + struct ZoO_knowledge k [const restrict static 1] +) +{ + ZoO_index word_id; + ZoO_char * stored_word; + + if (word_length == 0) + { + return 0; + } + + if + ( + ZoO_knowledge_learn_word + ( + k, + (string + word_start), + word_length, + &word_id + ) < 0 + ) + { + return -1; + } + + if (add_word_id_to_sequence(word_id, sequence, sequence_length) < 0) + { + return -1; + } + + return 0; +} + +static int add_finding_to_sequence +( + const ZoO_char string [const restrict static 1], + const ZoO_index word_start, + const ZoO_index word_length, + ZoO_index * sequence [const restrict static 1], + ZoO_index sequence_length [const restrict static 1], + struct ZoO_knowledge k [const restrict static 1] +) +{ + ZoO_index punctuation; + + if (word_is_punctuation_terminated(string, word_start, word_length)) + { + punctuation = 1; + } + else + { + punctuation = 0; + } + + if + ( + add_word_to_sequence + ( + string, + word_start, + (word_length - punctuation), + sequence, + sequence_length, + k + ) < 0 + ) + { + return -1; + } + + if + ( + (punctuation == 1) + && + ( + add_punctuation_to_sequence + ( + string, + string[word_start + word_length - 1], + sequence, + sequence_length, + k + ) < 0 + ) + ) + { + return -1; + } + + return 0; +} + +static int find_word +( + const ZoO_char string [const restrict static 1], + const ZoO_index string_length, + const ZoO_index offset, + ZoO_index word_start [const restrict static 1], + ZoO_index word_length [const restrict static 1] +) +{ + ZoO_index i; + + i = offset; + + while ((string[i] == ' ') && (i < string_length)) + { + i += 1; + } + + if (i >= string_length) + { + return -1; + } + + *word_start = i; + + while ((string[i] != ' ') && (i < string_length)) + { + i += 1; + } + + if (i >= string_length) + { + return -1; + } + + *word_length = (i - *word_start); + + return 0; +} + +/******************************************************************************/ +/** EXPORTED ******************************************************************/ +/******************************************************************************/ +int ZoO_sequence_from_undercase_string +( + const ZoO_char string [const restrict], + const ZoO_index string_length, + struct ZoO_knowledge k [const restrict static 1], + ZoO_index * sequence [const restrict static 1], + ZoO_index sequence_length [const restrict static 1] +) +{ + ZoO_index word_start, word_length; + ZoO_index i; + + i = 0; + + *sequence = (ZoO_index *) NULL; + *sequence_length = 0; + + if + ( + add_word_id_to_sequence + ( + ZoO_START_OF_SEQUENCE_ID, + sequence, + sequence_length + ) < 0 + ) + { + return -1; + } + + while (i < string_length) + { + if (find_word(string, i, string_length, &word_start, &word_length) < 0) + { + break; + } + + if + ( + add_finding_to_sequence + ( + string, + word_start, + word_length, + sequence, + sequence_length, + k + ) < 0 + ) + { + free((void *) *sequence); + *sequence = (ZoO_index *) NULL; + *sequence_length = 0; + + return -1; + } + + i = (word_start + word_length); + } + + if + ( + add_word_id_to_sequence + ( + ZoO_END_OF_SEQUENCE_ID, + sequence, + sequence_length + ) < 0 + ) + { + free((void *) *sequence); + + *sequence = (ZoO_index *) NULL; + *sequence_length = 0; + + return -1; + } + + return 0; +} diff --git a/src/core/sequence_types.h b/src/core/sequence_types.h index 717d418..c260a8a 100644 --- a/src/core/sequence_types.h +++ b/src/core/sequence_types.h @@ -3,7 +3,8 @@ #define ZoO_START_OF_SEQUENCE_ID 0 #define ZoO_END_OF_SEQUENCE_ID 1 +#define ZoO_ACTION_SEQUENCE_ID 2 -#define ZoO_RESERVED_IDS_COUNT 2 +#define ZoO_RESERVED_IDS_COUNT 3 #endif diff --git a/src/io/data_input.c b/src/file/data_input.c index e31d33b..e31d33b 100644 --- a/src/io/data_input.c +++ b/src/file/data_input.c diff --git a/src/io/data_input.h b/src/file/data_input.h index a2f004b..a2f004b 100644 --- a/src/io/data_input.h +++ b/src/file/data_input.h diff --git a/src/io/data_input_types.h b/src/file/data_input_types.h index bd2709b..bd2709b 100644 --- a/src/io/data_input_types.h +++ b/src/file/data_input_types.h diff --git a/src/io/data_output.c b/src/file/data_output.c index 796d3d0..04e3964 100644 --- a/src/io/data_output.c +++ b/src/file/data_output.c @@ -1,4 +1,5 @@ #define _POSIX_C_SOURCE 200809L + #include <stdlib.h> #include <string.h> #include <errno.h> diff --git a/src/io/data_output.h b/src/file/data_output.h index ef963a0..ef963a0 100644 --- a/src/io/data_output.h +++ b/src/file/data_output.h diff --git a/src/io/CMakeLists.txt b/src/io/CMakeLists.txt deleted file mode 100644 index c36413a..0000000 --- a/src/io/CMakeLists.txt +++ /dev/null @@ -1,9 +0,0 @@ -set( - SRC_FILES ${SRC_FILES} - ${CMAKE_CURRENT_SOURCE_DIR}/parameters.c - ${CMAKE_CURRENT_SOURCE_DIR}/network.c - ${CMAKE_CURRENT_SOURCE_DIR}/data_input.c - ${CMAKE_CURRENT_SOURCE_DIR}/data_output.c -) -set(SRC_FILES ${SRC_FILES} PARENT_SCOPE) - diff --git a/src/io/parameters_types.h b/src/io/parameters_types.h deleted file mode 100644 index 92a9e30..0000000 --- a/src/io/parameters_types.h +++ /dev/null @@ -1,21 +0,0 @@ -#ifndef _ZoO_IO_PARAMETERS_TYPES_H_ -#define _ZoO_IO_PARAMETERS_TYPES_H_ - -struct ZoO_parameters -{ - const char * restrict data_filename; - const char * restrict new_data_filename; - - const char * restrict irc_server_addr; - const char * restrict irc_server_port; - const char * restrict irc_server_channel; - const char * restrict irc_username; - const char * restrict irc_realname; - - int reply_rate; - - int aliases_count; - const char * restrict * restrict aliases; -}; - -#endif diff --git a/src/io/network.c b/src/irc/network.c index edafd4f..edafd4f 100644 --- a/src/io/network.c +++ b/src/irc/network.c diff --git a/src/io/network.h b/src/irc/network.h index 647b19c..647b19c 100644 --- a/src/io/network.h +++ b/src/irc/network.h diff --git a/src/io/network_types.h b/src/irc/network_types.h index 9a328a7..9a328a7 100644 --- a/src/io/network_types.h +++ b/src/irc/network_types.h diff --git a/src/knowledge/CMakeLists.txt b/src/knowledge/CMakeLists.txt new file mode 100644 index 0000000..1245321 --- /dev/null +++ b/src/knowledge/CMakeLists.txt @@ -0,0 +1,11 @@ +set( + SRC_FILES ${SRC_FILES} + ${CMAKE_CURRENT_SOURCE_DIR}/knowledge.c + ${CMAKE_CURRENT_SOURCE_DIR}/knowledge_finalize.c + ${CMAKE_CURRENT_SOURCE_DIR}/knowledge_learn_sequence.c + ${CMAKE_CURRENT_SOURCE_DIR}/knowledge_learn_word.c + ${CMAKE_CURRENT_SOURCE_DIR}/knowledge_search.c +) + +set(SRC_FILES ${SRC_FILES} PARENT_SCOPE) + diff --git a/src/core/knowledge.c b/src/knowledge/knowledge.c index 94d76cd..a72969e 100644 --- a/src/core/knowledge.c +++ b/src/knowledge/knowledge.c @@ -2,14 +2,20 @@ #include <string.h> #include <stdint.h> /* defines SIZE_MAX */ -#include "../io/error.h" +#include "../cli/cli.h" #include "knowledge.h" /** Basic functions of the ZoO_knowledge structure ****************************/ + +/* See: "knowledge.h" */ void ZoO_knowledge_initialize (struct ZoO_knowledge k [const static 1]) { k->words = (struct ZoO_knowledge_word *) NULL; k->words_length = 0; k->words_sorted = (ZoO_index *) NULL; + + k->sequences = (ZoO_index **) NULL; + k->sequences_length = 0; + k->sequences_sorted = (ZoO_index *) NULL; } diff --git a/src/core/knowledge.h b/src/knowledge/knowledge.h index 057e436..51d94c4 100644 --- a/src/core/knowledge.h +++ b/src/knowledge/knowledge.h @@ -1,5 +1,5 @@ -#ifndef _ZoO_CORE_KNOWLEDGE_H_ -#define _ZoO_CORE_KNOWLEDGE_H_ +#ifndef _ZoO_KNOWLEDGE_KNOWLEDGE_H_ +#define _ZoO_KNOWLEDGE_KNOWLEDGE_H_ #include "../core/char_types.h" #include "../core/index_types.h" @@ -10,11 +10,9 @@ void ZoO_knowledge_initialize (struct ZoO_knowledge k [const restrict static 1]) void ZoO_knowledge_finalize (struct ZoO_knowledge k [const restrict static 1]); - /* * When returning 0: - * {word} was either added to {k} or its representation in {k} has its - * occurrences count increased. + * {word} was added to {k}, or was already there. * {*result} indicates where {word} is in {k->words}. * * When returning -1: @@ -22,18 +20,28 @@ void ZoO_knowledge_finalize (struct ZoO_knowledge k [const restrict static 1]); * {k} remains semantically unchanged. * {*result} may or may not have been altered. */ -int ZoO_knowledge_learn +int ZoO_knowledge_learn_word ( struct ZoO_knowledge k [const static 1], const ZoO_char word [const restrict static 1], + const ZoO_index word_length, ZoO_index result [const restrict static 1] ); int ZoO_knowledge_learn_sequence ( - struct ZoO_knowledge k [const static 1], - const ZoO_index sequence [const restrict], - const ZoO_index sequence_length + struct ZoO_knowledge k [const restrict static 1], + const ZoO_index sequence [const restrict static 1], + const ZoO_index sequence_length, + const ZoO_index markov_order +); + +int ZoO_knowledge_learn_markov_sequence +( + struct ZoO_knowledge k [const restrict static 1], + const ZoO_index sequence [const restrict static 1], + const ZoO_index sequence_length, + const ZoO_index markov_order ); int ZoO_knowledge_get_following_sequences_ref @@ -74,6 +82,7 @@ int ZoO_knowledge_find_word_id ( const struct ZoO_knowledge k [const restrict static 1], const ZoO_char word [const restrict static 1], + const size_t word_size, ZoO_index result [const restrict static 1] ); diff --git a/src/core/knowledge_finalize.c b/src/knowledge/knowledge_finalize.c index e4deda6..36a7406 100644 --- a/src/core/knowledge_finalize.c +++ b/src/knowledge/knowledge_finalize.c @@ -2,11 +2,11 @@ #include <string.h> #include <stdint.h> /* defines SIZE_MAX */ -#include "../io/error.h" +#include "../cli/cli.h" #include "knowledge.h" -void knowledge_sequence_collection_finalize +static void knowledge_sequence_collection_finalize ( struct ZoO_knowledge_sequence_collection c [const restrict static 1] ) @@ -54,7 +54,7 @@ void knowledge_sequence_collection_finalize } } -void knowledge_word_finalize +static void knowledge_word_finalize ( struct ZoO_knowledge_word w [const restrict static 1] ) @@ -73,6 +73,7 @@ void knowledge_word_finalize knowledge_sequence_collection_finalize(&(w->preceded)); } +/* See: "knowledge.h" */ void ZoO_knowledge_finalize (struct ZoO_knowledge k [const restrict static 1]) { ZoO_index i; diff --git a/src/knowledge/knowledge_learn_sequence.c b/src/knowledge/knowledge_learn_sequence.c new file mode 100644 index 0000000..23a5ca7 --- /dev/null +++ b/src/knowledge/knowledge_learn_sequence.c @@ -0,0 +1,324 @@ +#include <stdlib.h> +#include <string.h> +#include <stdint.h> /* defines SIZE_MAX */ + +#include "../core/sequence.h" + +#include "../cli/cli.h" + +#include "knowledge.h" + +/******************************************************************************/ +/** INITIALIZE ****************************************************************/ +/******************************************************************************/ +static void set_nth_sequence +( + struct ZoO_knowledge k [const restrict static 1], + const ZoO_index sorted_sequence_id, + const ZoO_index sequence_id +) +{ + /* Safe: (> k->sequences_length 1) */ + if (sorted_sequence_id < (k->sequences_length - 1)) + { + memmove + ( + /* Safe: (=< (+ sorted_sequence_id 1) k->sequences_length) */ + (void *) (k->sequences_sorted + (sorted_sequence_id + 1)), + (const void *) (k->sequences_sorted + sorted_sequence_id), + ((k->sequences_length - 1) - sorted_sequence_id) + ); + } + + k->sequences_sorted[sorted_sequence_id] = sequence_id; +} + +/******************************************************************************/ +/** ALLOCATING MEMORY *********************************************************/ +/******************************************************************************/ +static int reallocate_sequences_list +( + struct ZoO_knowledge k [const restrict static 1] +) +{ + ZoO_index ** new_sequences; + + if ((SIZE_MAX / sizeof(ZoO_index *)) > (size_t) k->sequences_length) + { + ZoO_S_ERROR + ( + "Unable to store the size of the sequences list, as it would overflow" + "size_t variables." + ); + + return -1; + } + + new_sequences = + (ZoO_index **) realloc + ( + (void *) k->sequences, + (((size_t) k->sequences_length) * sizeof(ZoO_index *)) + ); + + if (new_sequences == (ZoO_index **) NULL) + { + ZoO_S_ERROR + ( + "Unable to allocate the memory required for the new sequence list." + ); + + return -1; + } + + k->sequences = new_sequences; + + return 0; +} + +static int reallocate_sequences_sorted_list +( + struct ZoO_knowledge k [const restrict static 1] +) +{ + ZoO_index * new_sequences_sorted; + + if ((SIZE_MAX / sizeof(ZoO_index)) > (size_t) k->sequences_length) + { + ZoO_S_ERROR + ( + "Unable to store the size of the sorted sequences list, as it would" + " overflow size_t variables." + ); + + return -1; + } + + new_sequences_sorted = + (ZoO_index *) realloc + ( + (void *) k->sequences_sorted, + ((size_t) k->sequences_length) * sizeof(ZoO_index) + ); + + if (new_sequences_sorted == (ZoO_index *) NULL) + { + ZoO_S_ERROR + ( + "Unable to allocate the memory required for the new sorted sequences" + " list." + ); + + return -1; + } + + k->sequences_sorted = new_sequences_sorted; + + return 0; +} + +/* Pre: (=< ZoO_INDEX_MAX SIZE_MAX) */ +static ZoO_index * copy_sequence +( + const ZoO_index base [const restrict static 1], + const ZoO_index base_length, + const ZoO_index markov_order +) +{ + ZoO_index * result; + + result = (ZoO_index *) calloc((size_t) base_length, sizeof(ZoO_index)); + + if (result == (ZoO_index *) NULL) + { + ZoO_S_ERROR + ( + "Unable to allocate the memory required to store a new sequence." + ); + + return (ZoO_index *) NULL; + } + + memcpy + ( + (void *) result, + (const void *) base, + (((size_t) base_length) * sizeof(ZoO_index)) + ); + + return result; +} + +static int add_sequence +( + struct ZoO_knowledge k [const restrict static 1], + const ZoO_index sequence [const restrict static 1], + const ZoO_index sequence_length, + const ZoO_index markov_order, /* Pre (> markov_order 1) */ + const ZoO_index sequence_id, + const ZoO_index sorted_sequence_id +) +{ + ZoO_index * stored_sequence; + + if (k->sequences_length == ZoO_INDEX_MAX) + { + ZoO_S_ERROR + ( + "Unable to add sequence: the variable that stores the number of known " + "sequences would overflow." + ); + + return -1; + } + + stored_sequence = copy_sequence(sequence, sequence_length, markov_order); + + if (stored_sequence == (ZoO_index *) NULL) + { + return -1; + } + + k->sequences_length += 1; + + if (reallocate_sequences_list(k) < 0) + { + k->sequences_length -= 1; + + return -1; + } + + k->sequences[sequence_id] = stored_sequence; + + if (reallocate_sequences_sorted_list(k) < 0) + { + k->sequences_length -= 1; + + return -1; + } + + set_nth_sequence(k, sorted_sequence_id, sequence_id); + + return -1; +} + +/******************************************************************************/ +/** SEARCH ********************************************************************/ +/******************************************************************************/ + +static int find_sequence +( + const struct ZoO_knowledge k [const static 1], + const ZoO_index sequence [const restrict static 1], + const ZoO_index sequence_length, + const ZoO_index markov_order, /* Pre: (> 1) */ + ZoO_index sequence_id [const restrict static 1] +) +{ + /* This is a binary search */ + int cmp; + ZoO_index i, current_min, current_max; + const ZoO_index markov_sequence_length = (markov_order - 1); + + /* Handles the case where the list is empty ********************************/ + current_max = k->sequences_length; + + if (current_max == 0) + { + *sequence_id = 0; + + return -1; + } + /***************************************************************************/ + + current_min = 0; + current_max -= 1; + + for (;;) + { + i = (current_min + ((current_max - current_min) / 2)); + + cmp = + ZoO_sequence_cmp + ( + k->sequences[k->sequences_sorted[i]], + markov_sequence_length, + sequence, + sequence_length + ); + + if (cmp > 0) + { + current_min = (i + 1); + + if (current_min > current_max) + { + *sequence_id = current_min; + + return -1; + } + } + else if (cmp < 0) + { + if ((current_min > current_max) || (i == 0)) + { + *sequence_id = i; + + return -1; + } + + current_max = (i - 1); + } + else + { + *sequence_id = k->sequences_sorted[i]; + + return 0; + } + } +} + +/******************************************************************************/ +/** EXPORTED ******************************************************************/ +/******************************************************************************/ + +int ZoO_knowledge_learn_markov_sequence +( + struct ZoO_knowledge k [const restrict static 1], + const ZoO_index sequence [const restrict static 1], + const ZoO_index sequence_length, + const ZoO_index markov_order, /* Pre (> markov_order 1) */ + ZoO_index sequence_id [const restrict static 1] +) +{ + ZoO_index sorted_id; + + if + ( + find_sequence + ( + k, + sequence, + sequence_length, + markov_order, + sequence_id + ) == 0 + ) + { + return 0; + } + + sorted_id = *sequence_id; + *sequence_id = k->sequences_length; + + return + add_sequence + ( + k, + sequence, + sequence_length, + markov_order, + *sequence_id, + sorted_id + ); +} diff --git a/src/knowledge/knowledge_learn_word.c b/src/knowledge/knowledge_learn_word.c new file mode 100644 index 0000000..f55ac5b --- /dev/null +++ b/src/knowledge/knowledge_learn_word.c @@ -0,0 +1,276 @@ +#include <stdlib.h> +#include <string.h> +#include <stdint.h> /* defines SIZE_MAX */ + +#include "../cli/cli.h" + +#include "knowledge.h" + +/******************************************************************************/ +/** INITIALIZING STRUCTURES ***************************************************/ +/******************************************************************************/ + +static void initialize_sequence_collection +( + struct ZoO_knowledge_sequence_collection c [const restrict static 1] +) +{ + c->sequences_ref = (ZoO_index *) NULL; + c->sequences_ref_length = 0; + c->sequences_ref_sorted = (ZoO_index *) NULL; + c->occurrences = (ZoO_index *) NULL; + c->targets = (ZoO_index **) NULL; + c->targets_length = (ZoO_index *) NULL; + c->targets_occurrences = (ZoO_index **) NULL; +} + +static void initialize_word +( + struct ZoO_knowledge_word w [const restrict static 1] +) +{ + w->word = (const ZoO_char *) NULL; + w->word_size = 0; + w->occurrences = 0; + + initialize_sequence_collection(&(w->followed)); + initialize_sequence_collection(&(w->preceded)); +} + +/******************************************************************************/ +/** ALLOCATING MEMORY *********************************************************/ +/******************************************************************************/ +static ZoO_char * copy_word +( + const ZoO_char original [const restrict static 1], + const ZoO_index original_length +) +{ + ZoO_char * result; + + result = + (ZoO_char *) + calloc + ( + (size_t) (original_length + 1), + sizeof(ZoO_char) + ); + + if (result == (ZoO_char *) NULL) + { + ZoO_S_ERROR("Unable to allocate memory to store new word."); + + return (ZoO_char *) NULL; + } + + memcpy + ( + (void *) result, + (const void *) original, + (((size_t) original_length) * sizeof(ZoO_char)) + ); + + result[original_length] = '\0'; + + return 0; +} + +static int reallocate_words_list +( + struct ZoO_knowledge k [const restrict static 1] +) +{ + struct ZoO_knowledge_word * new_words; + + if + ( + (SIZE_MAX / sizeof(struct ZoO_knowledge_word)) > (size_t) k->words_length + ) + { + ZoO_S_ERROR + ( + "Unable to store the size of the words list, as it would overflow" + "size_t variables." + ); + + return -1; + } + + new_words = + (struct ZoO_knowledge_word *) realloc + ( + (void *) k->words, + (((size_t) k->words_length) * sizeof(struct ZoO_knowledge_word)) + ); + + if (new_words == (struct ZoO_knowledge_word *) NULL) + { + ZoO_S_ERROR + ( + "Unable to allocate the memory required for the new words list." + ); + + return -1; + } + + k->words = new_words; + + return 0; +} + +static int reallocate_words_sorted_list +( + struct ZoO_knowledge k [const restrict static 1] +) +{ + ZoO_index * new_words_sorted; + + /* + * This has already been tested previously for a struct ZoO_knowledge_word, + * whose size is bigger than a ZoO_index. + * */ + /* + if ((SIZE_MAX / sizeof(ZoO_index)) > (size_t) k->words_length) + { + ZoO_S_ERROR + ( + "Unable to store the size of the sorted words list, as it would" + " overflow size_t variables." + ); + + return -1; + } + */ + + new_words_sorted = + (ZoO_index *) realloc + ( + (void *) k->words_sorted, + (((size_t) k->words_length) * sizeof(ZoO_index)) + ); + + if (new_words_sorted == (ZoO_index *) NULL) + { + ZoO_S_ERROR + ( + "Unable to allocate the memory required for the new sorted words list." + ); + + return -1; + } + + k->words_sorted = new_words_sorted; + + return 0; +} + +static void set_nth_word +( + struct ZoO_knowledge k [const restrict static 1], + const ZoO_index sorted_word_id, + const ZoO_index word_id +) +{ + /* Safe: (> k->words_length 1) */ + if (sorted_word_id < (k->words_length - 1)) + { + memmove + ( + /* Safe: (=< (+ sorted_word_id 1) k->words_length) */ + (void *) (k->words_sorted + (sorted_word_id + 1)), + (const void *) (k->words_sorted + sorted_word_id), + ((k->words_length - 1) - sorted_word_id) + ); + } + + k->words_sorted[sorted_word_id] = word_id; +} + +static int add_word +( + struct ZoO_knowledge k [const restrict static 1], + const ZoO_char word [const restrict static 1], + const ZoO_index word_length, + const ZoO_index word_id, + const ZoO_index sorted_word_id +) +{ + ZoO_char * stored_word; + + if (k->words_length == ZoO_INDEX_MAX) + { + ZoO_S_ERROR + ( + "Unable to add word: the variable that stores the number of known " + "words would overflow." + ); + + return -1; + } + + stored_word = copy_word(word, word_length); + + if (stored_word == (ZoO_char *) NULL) + { + return -1; + } + + k->words_length += 1; + + if (reallocate_words_list(k) < 0) + { + k->words_length -= 1; + + return -1; + } + + initialize_word(k->words + word_id); + + k->words[word_id].word = stored_word; + k->words[word_id].word_size = ((word_length + 1) * sizeof(ZoO_char)); + + if (reallocate_words_sorted_list(k) < 0) + { + k->words_length -= 1; + + return -1; + } + + set_nth_word(k, sorted_word_id, word_id); + + return -1; +} + +/******************************************************************************/ +/** EXPORTED ******************************************************************/ +/******************************************************************************/ + +int ZoO_knowledge_learn_word +( + struct ZoO_knowledge k [const restrict static 1], + const ZoO_char word [const restrict static 1], + const ZoO_index word_length, + ZoO_index word_id [const restrict static 1] +) +{ + ZoO_index sorted_id; + + if + ( + ZoO_knowledge_find_word_id + ( + k, + word, + (word_length * sizeof(ZoO_char)), + word_id + ) == 0 + ) + { + return 0; + } + + sorted_id = *word_id; + *word_id = k->words_length; + + return add_word(k, word, word_length, *word_id, sorted_id); +} diff --git a/src/core/knowledge_search.c b/src/knowledge/knowledge_search.c index d0c61ef..a48585b 100644 --- a/src/core/knowledge_search.c +++ b/src/knowledge/knowledge_search.c @@ -4,7 +4,7 @@ #include "../core/index.h" #include "../core/sequence.h" -#include "../io/error.h" +#include "../cli/cli.h" #include "knowledge.h" @@ -13,6 +13,7 @@ int ZoO_knowledge_find_word_id ( const struct ZoO_knowledge k [const restrict static 1], const ZoO_char word [const restrict static 1], + const size_t word_size, ZoO_index result [const restrict static 1] ) { @@ -39,7 +40,7 @@ int ZoO_knowledge_find_word_id { i = (current_min + ((current_max - current_min) / 2)); - cmp = ZoO_word_cmp(word, k->words[k->words_sorted[i]].word); + cmp = ZoO_word_cmp(word, word_size, k->words[k->words_sorted[i]].word); if (cmp > 0) { @@ -65,7 +66,7 @@ int ZoO_knowledge_find_word_id } else { - *result = i; + *result = k->words_sorted[i]; return 0; } diff --git a/src/core/knowledge_types.h b/src/knowledge/knowledge_types.h index acd239f..7eafc8b 100644 --- a/src/core/knowledge_types.h +++ b/src/knowledge/knowledge_types.h @@ -1,5 +1,5 @@ -#ifndef _ZoO_CORE_KNOWLEDGE_TYPES_H_ -#define _ZoO_CORE_KNOWLEDGE_TYPES_H_ +#ifndef _ZoO_KNOWLEDGE_KNOWLEDGE_TYPES_H_ +#define _ZoO_KNOWLEDGE_KNOWLEDGE_TYPES_H_ #include "../core/index_types.h" #include "../core/char_types.h" @@ -32,6 +32,7 @@ struct ZoO_knowledge ZoO_index ** sequences; ZoO_index sequences_length; ZoO_index * sequences_sorted; + ZoO_index sequences_length; }; #endif diff --git a/src/core/main.c b/src/main.c index bb4ae23..bb4ae23 100644 --- a/src/core/main.c +++ b/src/main.c diff --git a/src/pervasive.h b/src/pervasive.h index b830326..c7c53a2 100644 --- a/src/pervasive.h +++ b/src/pervasive.h @@ -3,6 +3,12 @@ #include <limits.h> +#define ZoO_DEBUG_ALL 1 + +#ifndef ZoO_DEBUG_ALL + #define ZoO_DEBUG_ALL 0 +#endif + #ifndef ZoO_NETWORK_TIMEOUT #define ZoO_NETWORK_TIMEOUT 200 #endif @@ -11,34 +17,6 @@ #define ZoO_MAX_REPLY_WORDS 64 #endif -#ifndef ZoO_DEFAULT_DATA_FILENAME - #define ZoO_DEFAULT_DATA_FILENAME "./memory.txt" -#endif - -#ifndef ZoO_DEFAULT_IRC_SERVER_ADDR - #define ZoO_DEFAULT_IRC_SERVER_ADDR "irc.foonetic.net" -#endif - -#ifndef ZoO_DEFAULT_IRC_SERVER_PORT - #define ZoO_DEFAULT_IRC_SERVER_PORT "6667" -#endif - -#ifndef ZoO_DEFAULT_IRC_SERVER_CHANNEL - #define ZoO_DEFAULT_IRC_SERVER_CHANNEL "#theborghivemind" -#endif - -#ifndef ZoO_DEFAULT_IRC_USERNAME - #define ZoO_DEFAULT_IRC_USERNAME "zeroofone" -#endif - -#ifndef ZoO_DEFAULT_IRC_REALNAME - #define ZoO_DEFAULT_IRC_REALNAME "Zero of One (bot)" -#endif - -#ifndef ZoO_DEFAULT_REPLY_RATE - #define ZoO_DEFAULT_REPLY_RATE 8 -#endif - #define ZoO__TO_STRING(x) #x #define ZoO_TO_STRING(x) ZoO__TO_STRING(x) #define ZoO_ISOLATE(a) do {a} while (0) |


