[numex] Separating rules from keys for Linux gcc compilation

scripts/geodata/i18n/numex.py

@@ -73,9 +73,10 @@ rule_type_map = {
     'ordinal_indicator': ORDINAL_INDICATOR,
 }
 
-numex_rule_template = u'{{"{key}", (numex_rule_t){{{left_context_type}, {right_context_type}, {rule_type}, {gender}, {category}, {radix}, {value}LL}}}}'
+numex_key_template = u'"{key}"'
+numex_rule_template = u'{{{left_context_type}, {right_context_type}, {rule_type}, {gender}, {category}, {radix}, {value}LL}}'
 
-stopword_rule_template = u'{{"{key}", NUMEX_STOPWORD_RULE}}'
+stopword_rule = u'NUMEX_STOPWORD_RULE'
 
 ordinal_indicator_template = u'{{"{key}", {gender}, {category}, "{value}"}}'
 
@@ -84,7 +85,11 @@ stopwords_template = u'"{word}"'
 language_template = u'{{"{language}", {whole_words_only}, {rule_index}, {num_rules}, {ordinal_indicator_index}, {num_ordinal_indicators}}}'
 
 numex_rules_data_template = u'''
-numex_rule_source_t numex_rules[] = {{
+char *numex_keys[] = {{
+    {numex_keys}
+}};
+
+numex_rule_t numex_rules[] = {{
     {numex_rules}
 }};
 
@@ -99,7 +104,9 @@ numex_language_source_t numex_languages[] = {{
 
 
 def parse_numex_rules(dirname=NUMEX_DATA_DIR, outfile=NUMEX_RULES_FILE):
+    all_keys = []
     all_rules = []
+
     all_ordinal_indicators = []
     all_stopwords = []
 
@@ -133,8 +140,8 @@ def parse_numex_rules(dirname=NUMEX_DATA_DIR, outfile=NUMEX_RULES_FILE):
             category = category_map[rule.get('category')]
             left_context_type = left_context_map[rule.get('left')]
             right_context_type = right_context_map[rule.get('right')]
+            all_keys.append(unicode(numex_key_template.format(key=key)))
             all_rules.append(unicode(numex_rule_template.format(
-                key=key,
                 language=language,
                 rule_type=rule_type,
                 gender=gender,
@@ -171,7 +178,8 @@ def parse_numex_rules(dirname=NUMEX_DATA_DIR, outfile=NUMEX_RULES_FILE):
         num_stopwords = len(stopwords)
 
         for stopword in stopwords:
-            all_rules.append(unicode(stopword_rule_template.format(key=stopword)))
+            all_keys.append(numex_key_template.format(key=unicode(stopword)))
+            all_rules.append(stopword_rule)
 
         num_rules = len(rules) + len(stopwords)
 
@@ -185,6 +193,8 @@ def parse_numex_rules(dirname=NUMEX_DATA_DIR, outfile=NUMEX_RULES_FILE):
         )))
 
     out.write(safe_encode(numex_rules_data_template.format(
+        numex_keys=u''',
+    '''.join(all_keys),
         numex_rules=u''',
     '''.join(all_rules),
         ordinal_indicator_rules=u''',

src/numex_rule.h

@@ -4,11 +4,6 @@
 #include <stdlib.h>
 #include "numex.h"
 
-typedef struct numex_rule_source {
-    char *key;
-    numex_rule_t rule;
-} numex_rule_source_t;
-
 typedef struct numex_language_source {
     char *name;
     bool whole_tokens_only;

src/numex_table_builder.c

@@ -36,7 +36,15 @@ int main(int argc, char **argv) {
 
     size_t num_languages = sizeof(numex_languages) / sizeof(numex_language_source_t);
 
-    size_t num_source_rules = sizeof(numex_rules) / sizeof(numex_rule_source_t);
+    size_t num_source_keys = sizeof(numex_keys) / sizeof(char *);
+    size_t num_source_rules = sizeof(numex_rules) / sizeof(numex_rule_t);
+
+    if (num_source_keys != num_source_rules) {
+        log_error("num_sourcE_keys != num_source_rules, aborting\n");
+        numex_module_teardown();
+        exit(1);
+    }
+
     size_t num_ordinal_indicator_rules = sizeof(ordinal_indicator_rules) / sizeof(ordinal_indicator_t);
 
     char_array *key = char_array_new();
@@ -60,15 +68,16 @@ int main(int argc, char **argv) {
         log_info("Doing language=%s\n", lang);
 
         for (j = rule_index; j < rule_index + num_rules; j++) {
-            numex_rule_source_t rule_source = numex_rules[j];
+            char *numex_key = numex_keys[j];
+            numex_rule_t rule = numex_rules[j];
 
-            value = rule_source.rule.rule_type != NUMEX_STOPWORD ? numex_table->rules->n : NUMEX_STOPWORD_INDEX;
-            numex_rule_array_push(numex_table->rules, rule_source.rule);
+            value = rule.rule_type != NUMEX_STOPWORD ? numex_table->rules->n : NUMEX_STOPWORD_INDEX;
+            numex_rule_array_push(numex_table->rules, rule);
 
             char_array_clear(key);
             char_array_cat(key, lang);
             char_array_cat(key, NAMESPACE_SEPARATOR_CHAR);
-            char_array_cat(key, rule_source.key);
+            char_array_cat(key, numex_key);
 
             char *str_key = char_array_get_string(key);