MeowOS-SDK/tuku/logic/reader/archive_source.c

/* logic/reader/archive_source.c — 档案册（CBZ/zip 当一本书）：页源契约的第二种实现。
 *
 * 契约、条目过滤与缓存纪律见 page_source.h / page_cache.h。实现要点：
 *   - 页表在 init 时一次建好（mz_open 只读 central directory；只看名字，不解压）；
 *   - open(i) 才解压单条目：mz_read_alloc → page_cache 落盘 → 注入的解码回调按路径解码。
 *     因此翻页/条漫两套消费代码与目录册完全一致（meta 走原始像素尺寸，open 走解码）；
 *   - meta(i) 只读该条目的**前 4KB**（mz_read_prefix）取图片头（不落缓存），结果缓存在
 *     meta_w/meta_h；M32 前是整条目解压，500 页排版 = 500 次全量解压；
 *   - 单个 mr_zip_book 的 mz_read* 由自带的互斥量串行化（解码线程与主线程 meta 会碰面）。
 *
 * 线程：count/page_name 只读页表，可任意线程；open 可由解码线程调用；meta 按 M6
 * 用法在主线程调用（条漫排版/当前页几何）。两个 mr_zip_book 实例互不影响，
 * 解压缓存（page_cache）是进程级共享的。
 */
#include "logic/reader/page_source.h"
#include "logic/reader/page_cache.h"
#include "logic/zip/zip.h"

#include <pthread.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/stat.h>

/* ============================================================ */
/* 条目过滤：图片后缀 + 路径安全 + 非隐藏 + 非目录               */
/* ============================================================ */
static int has_control(const char *name, unsigned len)
{
    for (unsigned i = 0; i < len; i++) {
        unsigned char c = (unsigned char)name[i];
        if (c < 0x20 || c == 0x7F) return 1;
    }
    return 0;
}

/* 逐段检查：拒绝 ".." 段与隐藏段（"._resource" 之类不参与成册）。 */
static int components_safe(const char *name)
{
    const char *p = name;
    while (*p) {
        const char *end = strchr(p, '/');
        size_t n = end ? (size_t)(end - p) : strlen(p);
        if (n == 2 && p[0] == '.' && p[1] == '.') return 0;
        if (n > 0 && p[0] == '.') return 0;
        if (!end) break;
        p = end + 1;
    }
    return 1;
}

int mr_zip_entry_is_page(const char *name, unsigned name_len)
{
    if (!name || name_len == 0 || name_len >= MR_ZIP_MAX_ENTRY_NAME) return 0;
    if (strlen(name) != name_len) return 0;              /* 内嵌 NUL：拒绝 */
    if (name[0] == '/' || name[0] == '\\') return 0;      /* 绝对路径 */
    if (name_len >= 2 && name[1] == ':') return 0;        /* Windows 盘符 */
    if (strchr(name, '\\')) return 0;                     /* 反斜杠分隔符：一律拒绝 */
    if (has_control(name, name_len)) return 0;
    if (name[name_len - 1] == '/') return 0;              /* 目录条目 */
    if (!components_safe(name)) return 0;
    return mr_path_is_image(name);
}

/* 页表排序：跨子目录按完整条目名自然序（中文按 UTF-8 字节序，数字段按数值）。 */
typedef struct page_ref {
    const char *name;
    int eidx;
} page_ref;

static int page_ref_cmp(const void *pa, const void *pb)
{
    const page_ref *a = (const page_ref *)pa;
    const page_ref *b = (const page_ref *)pb;
    int c = mr_natural_cmp(a->name, b->name);
    if (c != 0) return c;
    return a->eidx < b->eidx ? -1 : (a->eidx > b->eidx ? 1 : 0);
}

/* ============================================================ */
/* 生命周期                                                      */
/* ============================================================ */
static int entry_is_page(const mz_entry *e)
{
    if (!e || mz_entry_is_dir(e)) return 0;
    if (e->flags & 0x0001u) return 0;                     /* 加密条目：读不了，不列 */
    if (!mr_zip_entry_is_page(e->name, e->name_len)) return 0;
    if (e->uncomp_size == 0) return 0;
    if (e->uncomp_size > (uint64_t)MR_ZIP_PAGE_MAX_BYTES) return 0;   /* zip 炸弹 */
    return 1;
}

void mr_zip_book_free(mr_zip_book *zb)
{
    if (!zb) return;
    if (zb->arch) {
        mz_close((mz_archive *)zb->arch);
        zb->arch = NULL;
    }
    if (zb->lock) {
        pthread_mutex_destroy((pthread_mutex_t *)zb->lock);
        free(zb->lock);
        zb->lock = NULL;
    }
    free(zb->order);
    zb->order = NULL;
    free(zb->meta_w);
    zb->meta_w = NULL;
    free(zb->meta_h);
    zb->meta_h = NULL;
    zb->count = 0;
}

int mr_zip_book_init(mr_zip_book *zb, const char *path, int maxdim,
                     mr_book_load_fn load, mr_book_unload_fn unload, void *ud)
{
    if (!zb) return -1;
    memset(zb, 0, sizeof *zb);
    if (!path || !path[0]) return -1;
    if (strlen(path) >= sizeof zb->path) return -1;
    if (maxdim <= 0) maxdim = 720;

    mz_archive *z = NULL;
    if (mz_open(path, &z) != MZ_OK || !z) return -1;

    pthread_mutex_t *lk = (pthread_mutex_t *)malloc(sizeof *lk);
    if (!lk) {
        mz_close(z);
        return -1;
    }
    pthread_mutex_init(lk, NULL);

    snprintf(zb->path, sizeof zb->path, "%s", path);
    zb->arch = z;
    zb->lock = lk;
    zb->maxdim = maxdim;
    zb->load = load;
    zb->unload = unload;
    zb->ud = ud;

    struct stat st;
    if (stat(path, &st) == 0) {
        zb->mtime = (long long)st.st_mtime;
        zb->size = (long long)st.st_size;
    }

    int n = mz_count(z);
    page_ref *refs = NULL;
    int nref = 0, cap = 0;
    for (int i = 0; i < n && nref < MR_BOOK_MAX_PAGES; i++) {
        const mz_entry *e = mz_entry_at(z, i);
        if (!entry_is_page(e)) continue;
        if (nref >= cap) {
            int nc = cap ? cap * 2 : 32;
            page_ref *nr = (page_ref *)realloc(refs, (size_t)nc * sizeof *nr);
            if (!nr) {
                free(refs);
                mr_zip_book_free(zb);
                return -1;
            }
            refs = nr;
            cap = nc;
        }
        refs[nref].name = e->name;
        refs[nref].eidx = i;
        nref++;
    }
    if (nref > 1) qsort(refs, (size_t)nref, sizeof *refs, page_ref_cmp);
    if (nref > 0) {
        zb->order = (int *)malloc((size_t)nref * sizeof *zb->order);
        if (!zb->order) {
            free(refs);
            mr_zip_book_free(zb);
            return -1;
        }
        for (int i = 0; i < nref; i++) zb->order[i] = refs[i].eidx;
    }
    free(refs);
    zb->count = nref;
    return nref;
}

/* ============================================================ */
/* 页 → 缓存文件                                                 */
/* ============================================================ */
static const mz_entry *page_entry(mr_zip_book *zb, int index)
{
    if (!zb || !zb->arch || !zb->order) return NULL;
    if (index < 0 || index >= zb->count) return NULL;
    return mz_entry_at((mz_archive *)zb->arch, zb->order[index]);
}

int mr_zip_book_page_path(mr_zip_book *zb, int index, char *out, int outsz)
{
    if (!out || outsz <= 0) return -1;
    out[0] = '\0';

    const mz_entry *e = page_entry(zb, index);
    if (!e) return -1;

    char path[MR_CACHE_FILE_MAX];
    if (mr_cache_path(zb->path, zb->mtime, zb->size, e->name, index,
                      path, (int)sizeof path) != 0)
        return -1;
    if (mr_cache_touch(path)) {
        snprintf(out, (size_t)outsz, "%s", path);
        return 0;
    }

    pthread_mutex_lock((pthread_mutex_t *)zb->lock);
    void *buf = NULL;
    size_t len = 0;
    int rc = mz_read_alloc((mz_archive *)zb->arch, e, &buf, &len);
    pthread_mutex_unlock((pthread_mutex_t *)zb->lock);
    if (rc != MZ_OK || !buf) {
        free(buf);
        return -1;
    }
    int ok = mr_cache_store(path, buf, len);
    free(buf);
    if (!ok) return -1;
    snprintf(out, (size_t)outsz, "%s", path);
    return 0;
}

void mr_zip_book_page_name(const mr_zip_book *zb, int index, char *out, int outsz)
{
    if (!out || outsz <= 0) return;
    out[0] = '\0';
    const mz_entry *e = page_entry((mr_zip_book *)zb, index);
    if (!e) return;
    snprintf(out, (size_t)outsz, "%s", e->name);
}

/* ============================================================ */
/* 书库辅助：数页 / 取封面                                       */
/* ============================================================ */
int mr_zip_count_pages(const char *path)
{
    if (!path || !path[0]) return 0;
    mz_archive *z = NULL;
    if (mz_open(path, &z) != MZ_OK || !z) return 0;
    int n = mz_count(z), pages = 0;
    for (int i = 0; i < n && pages < MR_BOOK_MAX_PAGES; i++)
        if (entry_is_page(mz_entry_at(z, i))) pages++;
    mz_close(z);
    return pages;
}

int mr_zip_cover_path(const char *path, char *out, int outsz)
{
    if (!out || outsz <= 0) return 0;
    out[0] = '\0';
    if (!path || !path[0]) return 0;

    mr_zip_book zb;
    int n = mr_zip_book_init(&zb, path, 0, NULL, NULL, NULL);
    int ok = 0;
    if (n > 0) ok = (mr_zip_book_page_path(&zb, 0, out, outsz) == 0);
    mr_zip_book_free(&zb);
    return ok;
}

/* ============================================================ */
/* mr_book_ops 适配器（档案册）                                  */
/* ============================================================ */
static int zip_ops_count(void *ctx)
{
    mr_zip_book *zb = (mr_zip_book *)ctx;
    return zb ? zb->count : 0;
}

/* meta 只读条目**前缀**（≤4KB）取图片头（M32）：改前是 mz_read_alloc 整条目解压后再探
 * （500 页排版 = 500 次全量解压）。PNG IHDR 33 字节、JPEG SOF 在头几百字节、BMP 在
 * 54 字节 + 调色板内，4KB 足够；前缀探测失败且确实被截断时才回退整条目读（罕见/病态文件）。 */
#define MR_ZIP_META_PREFIX 4096

static int zip_meta_probe_prefix(mr_zip_book *zb, const mz_entry *e, int *w, int *h)
{
    unsigned char head[MR_ZIP_META_PREFIX];
    size_t head_len = 0;
    pthread_mutex_lock((pthread_mutex_t *)zb->lock);
    int rc = mz_read_prefix((mz_archive *)zb->arch, e, head, sizeof head, &head_len);
    pthread_mutex_unlock((pthread_mutex_t *)zb->lock);
    if (rc != MZ_OK) return 0;
    if (mr_probe_size_mem(head, head_len, w, h)) return 1;
    if (head_len < sizeof head) return 0;      /* 整个条目都读到了还探不出 → 坏头 */

    /* 前缀被截断且探不出尺寸（病态长的头部）：回退整条目读，保持旧语义。 */
    void *buf = NULL;
    size_t len = 0;
    pthread_mutex_lock((pthread_mutex_t *)zb->lock);
    rc = mz_read_alloc((mz_archive *)zb->arch, e, &buf, &len);
    pthread_mutex_unlock((pthread_mutex_t *)zb->lock);
    if (rc != MZ_OK || !buf) {
        free(buf);
        return 0;
    }
    int known = mr_probe_size_mem(buf, len, w, h);
    free(buf);
    return known;
}

/* meta 按需探测图片头；结果缓存，每页只读一次（前缀 4KB，不再整页解压）。主线程调用（M6 用法）。 */
static int zip_ops_meta(void *ctx, int index, int *w, int *h)
{
    mr_zip_book *zb = (mr_zip_book *)ctx;
    if (w) *w = 0;
    if (h) *h = 0;
    if (!zb || index < 0 || index >= zb->count) return 0;

    if (!zb->meta_w) {
        int *mw = (int *)calloc((size_t)zb->count, sizeof *mw);
        int *mh = (int *)calloc((size_t)zb->count, sizeof *mh);
        if (!mw || !mh) {
            free(mw);
            free(mh);
            return 0;
        }
        zb->meta_w = mw;
        zb->meta_h = mh;
    }
    if (zb->meta_w[index] > 0 && zb->meta_h[index] > 0) {
        if (w) *w = zb->meta_w[index];
        if (h) *h = zb->meta_h[index];
        return 1;
    }

    const mz_entry *e = page_entry(zb, index);
    if (!e) return 0;

    int pw = 0, ph = 0;
    if (!zip_meta_probe_prefix(zb, e, &pw, &ph)) return 0;

    zb->meta_w[index] = pw;
    zb->meta_h[index] = ph;
    if (w) *w = pw;
    if (h) *h = ph;
    return 1;
}

static meow_img *zip_ops_open(void *ctx, int index)
{
    mr_zip_book *zb = (mr_zip_book *)ctx;
    if (!zb || !zb->load) return NULL;
    char path[MR_CACHE_FILE_MAX];
    if (mr_zip_book_page_path(zb, index, path, (int)sizeof path) != 0) return NULL;
    return zb->load(path, zb->ud, zb->maxdim);
}

static void zip_ops_close(void *ctx, meow_img *img)
{
    mr_zip_book *zb = (mr_zip_book *)ctx;
    if (zb && zb->unload && img) zb->unload(img);
}

static void zip_ops_destroy(void *ctx)
{
    mr_zip_book *zb = (mr_zip_book *)ctx;
    if (zb) mr_zip_book_free(zb);
}

static const mr_book_ops g_zip_ops = {
    zip_ops_count, zip_ops_meta, zip_ops_open, zip_ops_close, zip_ops_destroy,
};

void mr_zip_book_bind(mr_book *b, mr_zip_book *zb)
{
    if (!b) return;
    b->ops = zb ? &g_zip_ops : NULL;
    b->ctx = zb;
}

/* ============================================================ */
/* 统一开册                                                      */
/* ============================================================ */
int mr_book_open_auto(mr_book *b, mr_book_handle *h, const char *path, int maxdim,
                      mr_book_load_fn load, mr_book_unload_fn unload, void *ud)
{
    if (b) {
        b->ops = NULL;
        b->ctx = NULL;
    }
    if (!b || !h) return -1;
    memset(h, 0, sizeof *h);
    if (!path || !path[0]) return -1;

    int n;
    /* 目录名也可能以 .cbz 结尾（罕见）：只有普通文件才当档案册。 */
    int is_archive = mr_path_is_archive(path);
    if (is_archive) {
        struct stat st;
        if (stat(path, &st) == 0 && S_ISDIR(st.st_mode)) is_archive = 0;
    }
    if (is_archive) {
        n = mr_zip_book_init(&h->u.zip, path, maxdim, load, unload, ud);
        if (n < 0) return -1;
        h->kind = MR_BOOK_KIND_ZIP;
        if (n > 0) mr_zip_book_bind(b, &h->u.zip);
    } else {
        n = mr_dir_book_init(&h->u.dir, path, maxdim, load, unload, ud);
        if (n < 0) return -1;
        h->kind = MR_BOOK_KIND_DIR;
        if (n > 0) mr_dir_book_bind(b, &h->u.dir);
    }
    return n;
}

void mr_book_close_auto(mr_book *b, mr_book_handle *h)
{
    if (b) mr_book_destroy(b);
    if (!h) return;
    if (h->kind == MR_BOOK_KIND_ZIP) mr_zip_book_free(&h->u.zip);
    else if (h->kind == MR_BOOK_KIND_DIR) mr_dir_book_free(&h->u.dir);
    memset(h, 0, sizeof *h);
}