/*
 * database_flatfile_parser.c
 * vim: expandtab:ts=4:sts=4:sw=4
 *
 * Copyright (C) 2026 Profanity Contributors
 *
 * This file is part of Profanity.
 *
 * Profanity is free software: you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation, either version 3 of the License, or
 * (at your option) any later version.
 *
 * Profanity is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with Profanity.  If not, see <https://www.gnu.org/licenses/>.
 *
 * Flat-file backend: type helpers, path helpers, escape/unescape,
 * line I/O, and the tolerant log-line parser.
 */

#include "config.h"

#include <sys/stat.h>
#include <glib.h>
#include <glib/gstdio.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <errno.h>

#include "log.h"
#include "common.h"
#include "config/files.h"
#include "database_flatfile.h"

// =========================================================================
// Type conversion helpers
// =========================================================================

const char*
ff_get_message_type_str(prof_msg_type_t type)
{
    switch (type) {
    case PROF_MSG_TYPE_CHAT:
        return "chat";
    case PROF_MSG_TYPE_MUC:
        return "muc";
    case PROF_MSG_TYPE_MUCPM:
        return "mucpm";
    case PROF_MSG_TYPE_UNINITIALIZED:
        return "chat";
    }
    return "chat";
}

prof_msg_type_t
ff_get_message_type_type(const char* const type)
{
    if (g_strcmp0(type, "chat") == 0) {
        return PROF_MSG_TYPE_CHAT;
    } else if (g_strcmp0(type, "muc") == 0) {
        return PROF_MSG_TYPE_MUC;
    } else if (g_strcmp0(type, "mucpm") == 0) {
        return PROF_MSG_TYPE_MUCPM;
    }
    return PROF_MSG_TYPE_CHAT;
}

const char*
ff_get_message_enc_str(prof_enc_t enc)
{
    switch (enc) {
    case PROF_MSG_ENC_OX:
        return "ox";
    case PROF_MSG_ENC_PGP:
        return "pgp";
    case PROF_MSG_ENC_OTR:
        return "otr";
    case PROF_MSG_ENC_OMEMO:
        return "omemo";
    case PROF_MSG_ENC_NONE:
        return "none";
    }
    return "none";
}

prof_enc_t
ff_get_message_enc_type(const char* const encstr)
{
    if (g_strcmp0(encstr, "ox") == 0) {
        return PROF_MSG_ENC_OX;
    } else if (g_strcmp0(encstr, "pgp") == 0) {
        return PROF_MSG_ENC_PGP;
    } else if (g_strcmp0(encstr, "otr") == 0) {
        return PROF_MSG_ENC_OTR;
    } else if (g_strcmp0(encstr, "omemo") == 0) {
        return PROF_MSG_ENC_OMEMO;
    }
    return PROF_MSG_ENC_NONE;
}

// =========================================================================
// Path helpers
// =========================================================================

// Sanitise a JID for use as a directory name.
// 1. Replace '@' with '_at_'
// 2. Reject / strip path-separator and traversal characters: '/', '\0', '..'
// This prevents a malicious federated JID like "../../../tmp/pwned" from
// escaping the log directory tree.
char*
ff_jid_to_dir(const char* jid)
{
    if (!jid || jid[0] == '\0')
        return NULL;

    // Replace '@' first
    char* step1 = str_replace(jid, "@", "_at_");
    if (!step1)
        return NULL;

    // Replace '/' and '\\' with '_' to prevent path traversal
    GString* out = g_string_sized_new(strlen(step1));
    for (const char* p = step1; *p; p++) {
        if (*p == '/' || *p == '\\') {
            g_string_append_c(out, '_');
        } else {
            g_string_append_c(out, *p);
        }
    }
    free(step1);

    // Collapse any remaining ".." sequences to "__" (belt-and-suspenders)
    char* result = g_string_free(out, FALSE);
    char* dotdot;
    while ((dotdot = strstr(result, "..")) != NULL) {
        dotdot[0] = '_';
        dotdot[1] = '_';
    }

    // Reject empty result
    if (result[0] == '\0') {
        g_free(result);
        return NULL;
    }
    return result;
}

// Get the base directory for a contact's logs:
// ~/.local/share/profanity/flatlog/{my_jid_dir}/{contact_jid_dir}/
char*
ff_get_contact_dir(const char* contact_barejid)
{
    if (!g_flatfile_account_jid || !contact_barejid)
        return NULL;

    auto_gchar gchar* data_path = files_get_data_path(DIR_FLATLOG);
    auto_gchar gchar* my_dir = ff_jid_to_dir(g_flatfile_account_jid);
    auto_gchar gchar* contact_dir = ff_jid_to_dir(contact_barejid);

    char* result = g_strdup_printf("%s/%s/%s", data_path, my_dir, contact_dir);
    return result;
}

// Get the log file path for a contact on a specific date:
// {contact_dir}/{YYYY_MM_DD}.log
char*
ff_get_log_path(const char* contact_barejid, GDateTime* dt)
{
    auto_gchar gchar* contact_dir = ff_get_contact_dir(contact_barejid);
    if (!contact_dir)
        return NULL;

    auto_gchar gchar* date_str = g_date_time_format(dt, "%Y_%m_%d");
    char* result = g_strdup_printf("%s/%s.log", contact_dir, date_str);
    return result;
}

// Ensure the directory exists, create if needed.
// Refuses to follow symlinks at the final component.
gboolean
ff_ensure_dir(const char* path)
{
    if (g_file_test(path, G_FILE_TEST_IS_DIR)) {
        // Verify it's not a symlink
        struct stat st;
        if (g_lstat(path, &st) == 0 && S_ISLNK(st.st_mode)) {
            log_error("flatfile: directory path is a symlink, refusing: %s", path);
            return FALSE;
        }
        return TRUE;
    }
    if (g_mkdir_with_parents(path, S_IRWXU) != 0) {
        log_error("flatfile: Could not create directory: %s (errno=%d)", path, errno);
        return FALSE;
    }
    return TRUE;
}

// =========================================================================
// Escape / unescape helpers
// =========================================================================
//
// Message text and metadata values from remote peers can contain arbitrary
// characters including newlines, pipes and brackets. Without escaping, a
// crafted message could inject fake log lines (log injection / format
// injection). We escape on write and unescape on read.

// Escape message body: \ -> \\, \n -> \n literal, \r -> \r literal
char*
ff_escape_message(const char* text)
{
    if (!text)
        return g_strdup("");
    GString* out = g_string_sized_new(strlen(text));
    for (const char* p = text; *p; p++) {
        switch (*p) {
        case '\\':
            g_string_append(out, "\\\\");
            break;
        case '\n':
            g_string_append(out, "\\n");
            break;
        case '\r':
            g_string_append(out, "\\r");
            break;
        default:
            g_string_append_c(out, *p);
            break;
        }
    }
    return g_string_free(out, FALSE);
}

// Unescape message body: \\  -> \, \n -> newline, \r -> CR
char*
ff_unescape_message(const char* text)
{
    if (!text)
        return g_strdup("");
    GString* out = g_string_sized_new(strlen(text));
    for (const char* p = text; *p; p++) {
        if (*p == '\\' && *(p + 1)) {
            p++;
            switch (*p) {
            case '\\':
                g_string_append_c(out, '\\');
                break;
            case 'n':
                g_string_append_c(out, '\n');
                break;
            case 'r':
                g_string_append_c(out, '\r');
                break;
            default:
                g_string_append_c(out, '\\');
                g_string_append_c(out, *p);
                break;
            }
        } else {
            g_string_append_c(out, *p);
        }
    }
    return g_string_free(out, FALSE);
}

// Escape metadata value (stanza_id, archive_id, replace_id):
// these come from remote servers and may contain |, ], \, newlines.
char*
ff_escape_meta_value(const char* val)
{
    if (!val || strlen(val) == 0)
        return NULL;
    GString* out = g_string_sized_new(strlen(val));
    for (const char* p = val; *p; p++) {
        switch (*p) {
        case '|':
            g_string_append(out, "\\|");
            break;
        case ']':
            g_string_append(out, "\\]");
            break;
        case '\\':
            g_string_append(out, "\\\\");
            break;
        case '\n':
            g_string_append(out, "\\n");
            break;
        case '\r':
            g_string_append(out, "\\r");
            break;
        default:
            g_string_append_c(out, *p);
            break;
        }
    }
    return g_string_free(out, FALSE);
}

// Unescape metadata value
char*
ff_unescape_meta_value(const char* val)
{
    if (!val)
        return NULL;
    GString* out = g_string_sized_new(strlen(val));
    for (const char* p = val; *p; p++) {
        if (*p == '\\' && *(p + 1)) {
            p++;
            switch (*p) {
            case '|':
                g_string_append_c(out, '|');
                break;
            case ']':
                g_string_append_c(out, ']');
                break;
            case '\\':
                g_string_append_c(out, '\\');
                break;
            case 'n':
                g_string_append_c(out, '\n');
                break;
            case 'r':
                g_string_append_c(out, '\r');
                break;
            default:
                g_string_append_c(out, '\\');
                g_string_append_c(out, *p);
                break;
            }
        } else {
            g_string_append_c(out, *p);
        }
    }
    return g_string_free(out, FALSE);
}

// =========================================================================
// Readline helper
// =========================================================================
//
// POSIX getline() for dynamic-length lines. Returns line without trailing
// newline, or NULL on EOF. Sets *truncated=TRUE if the line had no trailing
// newline at EOF (partial write detection).
char*
ff_readline(FILE* fp, gboolean* truncated)
{
    char* line = NULL;
    size_t cap = 0;
    ssize_t nread = getline(&line, &cap, fp);
    if (nread == -1) {
        free(line);
        return NULL;
    }
    // Guard against pathological lines that could exhaust memory.
    // getline() already allocated, so we free and skip if too long.
    if (nread > FF_MAX_LINE_LEN) {
        log_error("flatfile: line too long (%zd bytes), skipping", nread);
        // Check newline termination before freeing
        gboolean had_newline = (nread > 0 && line[nread - 1] == '\n');
        free(line);
        // Skip to next newline if the overlength line wasn't newline-terminated
        if (!had_newline) {
            int ch;
            while ((ch = fgetc(fp)) != EOF && ch != '\n') {}
        }
        if (truncated)
            *truncated = FALSE;
        // Return empty string so caller's loop continues (parse will reject it)
        return g_strdup("");
    }
    if (truncated)
        *truncated = FALSE;
    if (nread > 0 && line[nread - 1] == '\n') {
        line[--nread] = '\0';
    } else if (feof(fp)) {
        // Line without trailing newline at EOF — likely a partial write
        if (truncated)
            *truncated = TRUE;
    }
    return line;
}

// =========================================================================
// Line writer
// =========================================================================
//
// Format: {ISO8601} [{type}|{enc}|id:{stanza_id}|aid:{archive_id}|corrects:{replace_id}] {from_jid}/{resource}: {message}
//
// Message text is escaped: \\ -> \\\\, newline -> \\n, CR -> \\r
// Metadata values are escaped: |, ], \\, newline, CR
// Lines starting with '#' are comments.
// Empty lines are skipped.

void
ff_write_line(FILE* fp, const char* timestamp, const char* type, const char* enc,
              const char* stanza_id, const char* archive_id, const char* replace_id,
              const char* from_jid, const char* from_resource, const char* message_text)
{
    // Escape metadata values from remote peers
    auto_gchar gchar* safe_sid = ff_escape_meta_value(stanza_id);
    auto_gchar gchar* safe_aid = ff_escape_meta_value(archive_id);
    auto_gchar gchar* safe_rid = ff_escape_meta_value(replace_id);

    // Build metadata section: [type|enc|id:...|aid:...|corrects:...]
    GString* meta = g_string_new("[");
    g_string_append(meta, type ? type : "chat");
    g_string_append_c(meta, '|');
    g_string_append(meta, enc ? enc : "none");
    if (safe_sid) {
        g_string_append_printf(meta, "|id:%s", safe_sid);
    }
    if (safe_aid) {
        g_string_append_printf(meta, "|aid:%s", safe_aid);
    }
    if (safe_rid) {
        g_string_append_printf(meta, "|corrects:%s", safe_rid);
    }
    g_string_append_c(meta, ']');

    // Build sender — escape ": " in the resource part to prevent
    // the parser from splitting at the wrong point.
    GString* sender = g_string_new(from_jid ? from_jid : "unknown");
    if (from_resource && strlen(from_resource) > 0) {
        // Escape backslash and colon-space in resource
        GString* safe_res = g_string_sized_new(strlen(from_resource));
        for (const char* p = from_resource; *p; p++) {
            if (*p == '\\') {
                g_string_append(safe_res, "\\\\");
            } else if (*p == ':' && *(p + 1) == ' ') {
                g_string_append(safe_res, "\\: ");
                p++; // skip the space too
            } else {
                g_string_append_c(safe_res, *p);
            }
        }
        g_string_append_printf(sender, "/%s", safe_res->str);
        g_string_free(safe_res, TRUE);
    }

    // Escape message body to prevent log injection
    char* safe_msg = ff_escape_message(message_text);

    // Build complete line and write with a single fwrite()
    GString* full_line = g_string_new(NULL);
    g_string_printf(full_line, "%s %s %s: %s\n",
                    timestamp, meta->str, sender->str, safe_msg);

    size_t to_write = full_line->len;
    ssize_t written = fwrite(full_line->str, 1, to_write, fp);
    if (written != (ssize_t)to_write) {
        log_error("flatfile: partial write (%zd/%zu)", written, to_write);
    }

    g_string_free(full_line, TRUE);
    g_free(safe_msg);
    g_string_free(meta, TRUE);
    g_string_free(sender, TRUE);
}

// =========================================================================
// Parser helpers
// =========================================================================

// Find the first occurrence of 'ch' that is not preceded by an unescaped backslash.
const char*
ff_find_unescaped_char(const char* str, char ch)
{
    if (!str)
        return NULL;
    for (const char* p = str; *p; p++) {
        if (*p == '\\' && *(p + 1)) {
            p++; // skip escaped character
            continue;
        }
        if (*p == ch)
            return p;
    }
    return NULL;
}

// Split metadata content on unescaped '|'. Returns a NULL-terminated array.
// Caller must g_strfreev() the result.
char**
ff_split_meta(const char* meta)
{
    GPtrArray* arr = g_ptr_array_new();
    const char* start = meta;
    for (const char* p = meta;; p++) {
        if (*p == '\\' && *(p + 1)) {
            p++; // skip escaped char
            continue;
        }
        if (*p == '|' || *p == '\0') {
            g_ptr_array_add(arr, g_strndup(start, p - start));
            if (*p == '\0')
                break;
            start = p + 1;
        }
    }
    g_ptr_array_add(arr, NULL);
    return (char**)g_ptr_array_free(arr, FALSE);
}

// Find first unescaped ": " (colon-space) in a string.
const char*
ff_find_unescaped_colonspace(const char* str)
{
    if (!str)
        return NULL;
    for (const char* p = str; *p; p++) {
        if (*p == '\\' && *(p + 1)) {
            p++; // skip escaped character
            continue;
        }
        if (*p == ':' && *(p + 1) == ' ') {
            return p;
        }
    }
    return NULL;
}

// Unescape a sender resource: "\\" -> '\', "\: " -> ": "
char*
ff_unescape_sender_resource(const char* res)
{
    if (!res)
        return NULL;
    GString* out = g_string_sized_new(strlen(res));
    for (const char* p = res; *p; p++) {
        if (*p == '\\' && *(p + 1)) {
            p++;
            if (*p == '\\') {
                g_string_append_c(out, '\\');
            } else if (*p == ':' && *(p + 1) == ' ') {
                g_string_append(out, ": ");
                p++; // skip the space
            } else {
                // Unknown escape — preserve literally
                g_string_append_c(out, '\\');
                g_string_append_c(out, *p);
            }
        } else {
            g_string_append_c(out, *p);
        }
    }
    return g_string_free(out, FALSE);
}

// =========================================================================
// Line parser
// =========================================================================

void
ff_parsed_line_free(ff_parsed_line_t* pl)
{
    if (!pl)
        return;
    g_free(pl->timestamp_str);
    if (pl->timestamp)
        g_date_time_unref(pl->timestamp);
    g_free(pl->type);
    g_free(pl->enc);
    g_free(pl->stanza_id);
    g_free(pl->archive_id);
    g_free(pl->replace_id);
    g_free(pl->from_jid);
    g_free(pl->from_resource);
    g_free(pl->message);
    g_free(pl);
}

// Parse a single line. Returns NULL on parse failure.
// Line format: {timestamp} [{metadata}] {sender}: {message}
ff_parsed_line_t*
ff_parse_line(const char* line)
{
    if (!line || line[0] == '\0' || line[0] == '#') {
        return NULL;
    }

    // Strip trailing \r if present (CRLF handling)
    char* work = g_strdup(line);
    gsize len = strlen(work);
    if (len > 0 && work[len - 1] == '\r') {
        work[len - 1] = '\0';
        len--;
    }
    if (len == 0) {
        g_free(work);
        return NULL;
    }

    // UTF-8 validation
    const gchar* end;
    if (!g_utf8_validate(work, -1, &end)) {
        log_warning("flatfile: invalid UTF-8 at byte offset %ld", (long)(end - work));
        // Attempt Latin-1 fallback
        gsize br, bw;
        GError* err = NULL;
        char* converted = g_convert(work, -1, "UTF-8", "ISO-8859-1", &br, &bw, &err);
        if (converted) {
            g_free(work);
            work = converted;
        } else {
            if (err)
                g_error_free(err);
            g_free(work);
            return NULL;
        }
    }

    ff_parsed_line_t* result = g_malloc0(sizeof(ff_parsed_line_t));

    // Parse timestamp — everything up to first space followed by '['
    char* bracket_start = strchr(work, '[');
    char* first_space = strchr(work, ' ');

    if (bracket_start && first_space && first_space < bracket_start) {
        // Standard format with metadata: {timestamp} [{meta}] {sender}: {msg}
        result->timestamp_str = g_strndup(work, first_space - work);

        // Parse metadata section [...]
        const char* bracket_end = ff_find_unescaped_char(bracket_start + 1, ']');
        if (bracket_end) {
            char* meta_content = g_strndup(bracket_start + 1, bracket_end - bracket_start - 1);

            // Split by unescaped '|'
            char** parts = ff_split_meta(meta_content);
            if (parts) {
                int i = 0;
                for (; parts[i]; i++) {
                    if (i == 0) {
                        result->type = g_strdup(parts[i]);
                    } else if (i == 1) {
                        result->enc = g_strdup(parts[i]);
                    } else if (g_str_has_prefix(parts[i], "id:")) {
                        result->stanza_id = ff_unescape_meta_value(parts[i] + 3);
                    } else if (g_str_has_prefix(parts[i], "aid:")) {
                        result->archive_id = ff_unescape_meta_value(parts[i] + 4);
                    } else if (g_str_has_prefix(parts[i], "corrects:")) {
                        result->replace_id = ff_unescape_meta_value(parts[i] + 9);
                    }
                }
                g_strfreev(parts);
            }
            g_free(meta_content);

            // Parse sender: message after '] '
            const char* after_meta = bracket_end + 1;
            if (*after_meta == ' ')
                after_meta++;

            // Find first *unescaped* ': ' which separates sender from message.
            const char* colon = ff_find_unescaped_colonspace(after_meta);
            if (colon) {
                char* raw_sender = g_strndup(after_meta, colon - after_meta);

                // Split sender into jid/resource, then unescape resource
                char* slash = strchr(raw_sender, '/');
                if (slash) {
                    result->from_jid = g_strndup(raw_sender, slash - raw_sender);
                    result->from_resource = ff_unescape_sender_resource(slash + 1);
                } else {
                    result->from_jid = g_strdup(raw_sender);
                }
                g_free(raw_sender);

                result->message = ff_unescape_message(colon + 2);
            } else {
                // No ': ' found, treat entire rest as message with unknown sender
                result->from_jid = g_strdup("unknown");
                result->message = ff_unescape_message(after_meta);
            }
        } else {
            // No closing bracket — malformed metadata
            ff_parsed_line_free(result);
            g_free(work);
            return NULL;
        }
    } else if (first_space) {
        // Legacy/simple format without metadata: {timestamp} - {sender}: {msg}
        result->timestamp_str = g_strndup(work, first_space - work);
        result->type = g_strdup("chat");
        result->enc = g_strdup("none");

        char* rest = first_space + 1;
        // Skip " - " if present (chatlog.c format)
        if (g_str_has_prefix(rest, "- ")) {
            rest += 2;
        }

        char* colon = strstr(rest, ": ");
        if (colon) {
            char* sender = g_strndup(rest, colon - rest);
            char* slash = strchr(sender, '/');
            if (slash) {
                result->from_jid = g_strndup(sender, slash - sender);
                result->from_resource = g_strdup(slash + 1);
            } else {
                result->from_jid = g_strdup(sender);
            }
            g_free(sender);
            result->message = ff_unescape_message(colon + 2);
        } else {
            result->from_jid = g_strdup("unknown");
            result->message = ff_unescape_message(rest);
        }
    } else {
        // No space at all — can't parse
        ff_parsed_line_free(result);
        g_free(work);
        return NULL;
    }

    // Parse timestamp
    result->timestamp = g_date_time_new_from_iso8601(result->timestamp_str, NULL);
    if (!result->timestamp) {
        log_warning("flatfile: unparsable timestamp: %s", result->timestamp_str);
        ff_parsed_line_free(result);
        g_free(work);
        return NULL;
    }

    // Default type/enc if missing
    if (!result->type)
        result->type = g_strdup("chat");
    if (!result->enc)
        result->enc = g_strdup("none");

    g_free(work);
    return result;
}

// Convert parsed line to ProfMessage
ProfMessage*
ff_parsed_to_profmessage(ff_parsed_line_t* pl)
{
    ProfMessage* msg = message_init();
    msg->id = pl->stanza_id ? g_strdup(pl->stanza_id) : NULL;
    msg->from_jid = jid_create_from_bare_and_resource(pl->from_jid, pl->from_resource);
    msg->plain = g_strdup(pl->message ? pl->message : "");
    msg->timestamp = g_date_time_ref(pl->timestamp);
    msg->type = ff_get_message_type_type(pl->type);
    msg->enc = ff_get_message_enc_type(pl->enc);
    return msg;
}