Initial commit: Padre Matos Soares (1932) Bible corpus and pms CLIHEADmaster

1 files changed, 142 insertions, 0 deletions

diff --git a/src/pms_match.c b/src/pms_match.c
new file mode 100644
index 0000000..8140dd5
--- /dev/null
+++ b/src/pms_match.c
@@ -0,0 +1,142 @@
+#include "pms_match.h"
+#include "pms_data.h"
+#include "pms_util.h"
+#include <limits.h>
+#include <stdbool.h>
+#include <stddef.h>
+
+static bool verse_matches(const pms_ref *ref, const pms_verse *v) {
+    switch (ref->type) {
+    case PMS_REF_SEARCH: {
+        if (ref->book != 0 && (unsigned int)v->book != ref->book) {
+            return false;
+        }
+        if (ref->chapter != 0 && (unsigned int)v->chapter != ref->chapter) {
+            return false;
+        }
+        char folded[4096];
+        pms_fold_utf8(v->text, folded, sizeof(folded), NULL);
+        return regexec(&ref->search, folded, 0, NULL, 0) == 0;
+    }
+
+    case PMS_REF_EXACT:
+        return (unsigned int)v->book == ref->book &&
+               (ref->chapter == 0 || (unsigned int)v->chapter == ref->chapter) &&
+               (ref->verse == 0 || (unsigned int)v->verse == ref->verse);
+
+    case PMS_REF_EXACT_SET:
+        if ((unsigned int)v->book != ref->book || (unsigned int)v->chapter != ref->chapter) {
+            return false;
+        }
+        for (int i = 0; i < ref->verse_set_count; i++) {
+            if ((unsigned int)v->verse == ref->verse_set[i]) {
+                return true;
+            }
+        }
+        return false;
+
+    case PMS_REF_RANGE:
+        return (unsigned int)v->book == ref->book &&
+               ((ref->chapter_end == 0 && (unsigned int)v->chapter == ref->chapter) ||
+                ((unsigned int)v->chapter >= ref->chapter && (unsigned int)v->chapter <= ref->chapter_end)) &&
+               (ref->verse == 0 || (unsigned int)v->verse >= ref->verse) &&
+               (ref->verse_end == 0 || (unsigned int)v->verse <= ref->verse_end);
+
+    case PMS_REF_RANGE_EXT:
+        return (unsigned int)v->book == ref->book &&
+               (((unsigned int)v->chapter == ref->chapter && (unsigned int)v->verse >= ref->verse && ref->chapter != ref->chapter_end) ||
+                ((unsigned int)v->chapter > ref->chapter && (unsigned int)v->chapter < ref->chapter_end) ||
+                ((unsigned int)v->chapter == ref->chapter_end && (unsigned int)v->verse <= ref->verse_end && ref->chapter != ref->chapter_end) ||
+                (ref->chapter == ref->chapter_end && (unsigned int)v->chapter == ref->chapter &&
+                 (unsigned int)v->verse >= ref->verse && (unsigned int)v->verse <= ref->verse_end));
+
+    default:
+        return false;
+    }
+}
+
+static pms_range chapter_bounds(int i, const pms_config *config) {
+    pms_range bounds = {.start = i, .end = i + 1};
+    int book = pms_verses[i].book;
+    int chapter = pms_verses[i].chapter;
+
+    int max_before = config->chapter_context ? INT_MAX : config->before_context;
+    for (int n = 0; n < max_before && bounds.start > 0; n++) {
+        if (pms_verses[bounds.start - 1].book == book &&
+            (!config->chapter_context || pms_verses[bounds.start - 1].chapter == chapter)) {
+            bounds.start--;
+        } else {
+            break;
+        }
+    }
+
+    int max_after = config->chapter_context ? INT_MAX : config->after_context;
+    for (int n = 0; n < max_after && bounds.end < pms_verses_length; n++) {
+        if (pms_verses[bounds.end].book == book &&
+            (!config->chapter_context || pms_verses[bounds.end].chapter == chapter)) {
+            bounds.end++;
+        } else {
+            break;
+        }
+    }
+
+    return bounds;
+}
+
+static int find_next_match(const pms_ref *ref, const pms_config *config, int i, pms_range *range) {
+    for (; i < pms_verses_length; i++) {
+        if (verse_matches(ref, &pms_verses[i])) {
+            *range = chapter_bounds(i, config);
+            return i;
+        }
+    }
+    return -1;
+}
+
+int pms_next_verse(const pms_ref *ref, const pms_config *config, pms_match_data *data) {
+    if (data->current >= pms_verses_length) {
+        return -1;
+    }
+
+    if (data->current == 0 && data->next_match == 0) {
+        data->matches[0] = (pms_range){-1, -1};
+        data->matches[1] = (pms_range){-1, -1};
+
+        data->next_match = find_next_match(ref, config, 0, &data->matches[0]);
+        if (data->next_match < 0) {
+            data->current = pms_verses_length;
+            return -1;
+        }
+        data->next_match = find_next_match(ref, config, data->next_match + 1, &data->matches[1]);
+    }
+
+    while (data->matches[1].start != -1 && data->matches[1].start <= data->matches[0].end) {
+        if (data->matches[1].end > data->matches[0].end) {
+            data->matches[0].end = data->matches[1].end;
+        }
+        data->matches[1] = (pms_range){-1, -1};
+        if (data->next_match >= 0) {
+            data->next_match = find_next_match(ref, config, data->next_match + 1, &data->matches[1]);
+        }
+    }
+
+    if (data->current < data->matches[0].start) {
+        data->current = data->matches[0].start;
+    }
+
+    if (data->current < data->matches[0].end) {
+        return data->current++;
+    }
+
+    if (data->matches[1].start != -1) {
+        data->matches[0] = data->matches[1];
+        data->matches[1] = (pms_range){-1, -1};
+        if (data->next_match >= 0) {
+            data->next_match = find_next_match(ref, config, data->next_match + 1, &data->matches[1]);
+        }
+        return pms_next_verse(ref, config, data);
+    }
+
+    data->current = pms_verses_length;
+    return -1;
+}