src/context_store.c

/* context_store.c — Universal pattern-triggered context injection.
 *
 *   The regex-triggered skill loading pattern generalizes: ANY context
 *   (skills, tool docs, error recovery, codebase summaries, few-shot examples)
 *   is a (trigger, text) pair. The store matches all triggers against the
 *   current input and returns matching entries in priority order.
 *
 *   This replaces static Pri0 with a dynamic lookup: instead of stuffing the
 *   system prompt with all possible instructions, only inject what matches
 *   the current request. Information capacity goes from "fits in system
 *   prompt" to "fits on disk."
 *
 *   Triggers support: exact match, substring, regex (via callback if linked
 *   with PCRE2; otherwise substring fallback), and keyword intersection.
 *
 *   Compile:
 *     cc -O3 -std=c11 -fPIC -shared context_store.c -o libcontext_store.so
 */
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <stdint.h>
#include <ctype.h>

#define MAX_ENTRIES   128
#define MAX_TEXT_LEN  2048
#define MAX_TRIG_LEN  256
#define MATCH_EXACT   0
#define MATCH_SUBSTR  1
#define MATCH_KEYWORD 2

typedef struct {
    char id[32];
    char trigger[MAX_TRIG_LEN];
    char text[MAX_TEXT_LEN];
    int  match_type;   /* MATCH_EXACT, MATCH_SUBSTR, MATCH_KEYWORD */
    int  priority;     /* 0=highest (error recovery), 9=lowest (general knowledge) */
    int  cooldown;     /* skip this entry for N more calls after a hit */
} entry_t;

static entry_t g_store[MAX_ENTRIES];
static int     g_count = 0;
static int     g_hit_counts[MAX_ENTRIES]; /* for cooldown tracking */

/* ---- Public API ---- */

/* Add or update an entry. Returns index or -1 if store full. */
int store_add(const char *id, const char *trigger, const char *text,
              int match_type, int priority) {
    /* Update existing */
    for (int i = 0; i < g_count; i++) {
        if (strcmp(g_store[i].id, id) == 0) {
            strncpy(g_store[i].trigger, trigger, MAX_TRIG_LEN-1);
            strncpy(g_store[i].text, text, MAX_TEXT_LEN-1);
            g_store[i].match_type = match_type;
            g_store[i].priority = priority;
            return i;
        }
    }
    if (g_count >= MAX_ENTRIES) return -1;
    int i = g_count++;
    strncpy(g_store[i].id, id, 31);
    strncpy(g_store[i].trigger, trigger, MAX_TRIG_LEN-1);
    strncpy(g_store[i].text, text, MAX_TEXT_LEN-1);
    g_store[i].match_type = match_type;
    g_store[i].priority = priority;
    g_store[i].cooldown = 0;
    g_hit_counts[i] = 0;
    return i;
}

void store_remove(const char *id) {
    for (int i = 0; i < g_count; i++) {
        if (strcmp(g_store[i].id, id) == 0) {
            /* Shift remaining entries down */
            memmove(&g_store[i], &g_store[i+1], (g_count - i - 1) * sizeof(entry_t));
            memmove(&g_hit_counts[i], &g_hit_counts[i+1], (g_count - i - 1) * sizeof(int));
            g_count--;
            return;
        }
    }
}

/* Case-insensitive substring match */
static int ci_substr(const char *haystack, const char *needle) {
    if (!*needle) return 1;
    while (*haystack) {
        const char *h = haystack, *n = needle;
        while (*h && *n && tolower(*h) == tolower(*n)) { h++; n++; }
        if (!*n) return 1;
        haystack++;
    }
    return 0;
}

/* Check if a word from needle appears in haystack */
static int keyword_match(const char *haystack, const char *trigger) {
    char word[64];
    const char *p = trigger;
    while (*p) {
        while (*p == ' ' || *p == ',') p++;
        if (!*p) break;
        int len = 0;
        while (p[len] && p[len] != ' ' && p[len] != ',' && len < 63) len++;
        memcpy(word, p, len); word[len] = '\0';
        p += len;
        /* Check if this word appears in haystack */
        const char *h = haystack;
        while (*h) {
            while (*h == ' ') h++;
            const char *w = word;
            const char *s = h;
            while (*s && *w && tolower(*s) == tolower(*w)) { s++; w++; }
            if (!*w && (!*s || *s == ' ' || *s == '\n')) return 1;
            while (*h && *h != ' ') h++;
        }
    }
    return 0;
}

/* Match an entry against input. Returns 1 if match. */
static int entry_matches(const entry_t *e, const char *input) {
    switch (e->match_type) {
        case MATCH_EXACT:   return strcmp(input, e->trigger) == 0;
        case MATCH_SUBSTR:  return ci_substr(input, e->trigger);
        case MATCH_KEYWORD: return keyword_match(input, e->trigger);
        default:            return 0;
    }
}

/* Query the store. Returns matching entries in priority order as a
 * concatenated string. max_tokens limits total output (est. 4 chars/tok).
 * excluded_ids is a | separated list of ids to skip (already active).
 * Copies result to buf (caller-owned, at least bufsz bytes). */
int store_query(const char *input, int max_tokens, const char *excluded_ids,
                char *buf, int bufsz) {
    int budget = max_tokens * 4;
    int total = 0;
    int included = 0;

    buf[0] = '\0';

    /* Collect matching indices, sorted by priority */
    int matches[MAX_ENTRIES];
    int nm = 0;
    for (int i = 0; i < g_count; i++) {
        if (g_store[i].cooldown > 0) { g_store[i].cooldown--; continue; }
        if (!entry_matches(&g_store[i], input)) continue;
        /* Check exclusion */
        if (excluded_ids && *excluded_ids) {
            char ex_copy[512];
            strncpy(ex_copy, excluded_ids, 511);
            char *tok = strtok(ex_copy, "|");
            int skip = 0;
            while (tok) {
                if (strcmp(tok, g_store[i].id) == 0) { skip = 1; break; }
                tok = strtok(NULL, "|");
            }
            if (skip) continue;
        }
        /* Insert sorted by priority */
        int j;
        for (j = nm; j > 0 && g_store[matches[j-1]].priority > g_store[i].priority; j--)
            matches[j] = matches[j-1];
        matches[j] = i;
        nm++;
    }

    /* Write matching entries */
    for (int m = 0; m < nm; m++) {
        int i = matches[m];
        int len = strlen(g_store[i].text);
        if (total + len + 2 > budget && included > 0) break;
        int to_write = len;
        if (total + to_write + 2 > budget) to_write = budget - total - 2;
        if (to_write <= 0) break;

        if (included > 0) { buf[total++] = '\n'; buf[total] = '\0'; }
        memcpy(buf + total, g_store[i].text, to_write);
        total += to_write;
        buf[total] = '\0';
        included++;

        /* Set cooldown to prevent re-triggering */
        g_store[i].cooldown = 3;
        g_hit_counts[i]++;
    }

    return included;
}

/* Get store stats */
int store_count(void) { return g_count; }

/* Seed the store with default entries. Call once at startup. */
void store_init_defaults(void) {
    store_add("default-terminate",
        "quit exit stop shutdown",
        "To exit: type /exit or say 'quit'.",
        MATCH_KEYWORD, 9);

    store_add("error-oom",
        "out of memory OOM killed memory error allocation failed",
        "MEMORY FULL. Free RAM by: closing unused programs, reducing batch size, or using a smaller model. Current model: Hermes 3B Q4.",
        MATCH_KEYWORD, 0);

    store_add("error-timeout",
        "timed out timeout hang frozen stuck",
        "TIMEOUT DETECTED. The last operation took too long. Try: --no-mmap for faster loading, reduce max_tokens, or simplify the request.",
        MATCH_KEYWORD, 0);

    store_add("error-syntax",
        "syntax error compilation error parse error unexpected token",
        "SYNTAX ERROR. Check: missing semicolons, unmatched braces, wrong variable types. Compile with -Wall to see all warnings.",
        MATCH_KEYWORD, 1);

    store_add("tool-terminal",
        "terminal shell bash command cmd execute run",
        "TERMINAL: use terminal('command') tool. Always quote paths with spaces. Set timeout for long operations. Check exit code before proceeding.",
        MATCH_KEYWORD, 3);

    store_add("tool-file",
        "file read write save open edit",
        "FILES: use read_file(path) and write_file(path,content). Never write outside the project directory. Check file exists before reading.",
        MATCH_KEYWORD, 3);

    store_add("tool-git",
        "git commit push pull branch merge rebase",
        "GIT: use terminal with git commands. NEVER force push to main. Create feature branches. Commit messages should be descriptive.",
        MATCH_KEYWORD, 3);

    store_add("pattern-search",
        "search find grep locate look for",
        "SEARCH: use search_files(pattern) for content, or terminal('find ...') for filenames. Be specific in patterns to reduce noise.",
        MATCH_KEYWORD, 4);
}