perf(reader): M32 书架索引落盘 — 冷启动只 stat 434 个目录 mtime,跳过 readdir(3.0s→~0.15s)

M29 只解决了「回书架不重扫」;首次进书架仍要全量扫 434 目录(真机 3.0–3.4s,
其中 opendir+readdir 占 96%)。本单把扫描结果 + 每个扫描目录的 mtime 存到
MeowData/os.meow.reader/index/shelf.idx,冷启动逐目录 stat 比对 mtime,命中即
完全跳过 readdir;任何异常一律 fail-open 全量重扫。

先量再选(任务 1):宿主 434 目录/595 本/51 rar 虚拟库(贴真机量级)
- 434 次目录 stat = 0.48ms/pass(1.1µs/次,不含 readdir);
- 读 72KB 索引 = 0.009ms;全量扫描 3.54ms;索引命中 1.29ms(ts_fn=NULL)。
真机按 M28 实测 1079 次 stat=160–170ms(≈150µs/次)外推:434 次 ≈ 65ms,
远低于「可能不可行」的 1s 阈值 → 选「校验全部扫描目录 mtime」(非只 level-1)。
strace 断言:indexonly 载入 0 次 getdents64、恰好 434 次 stat(真的没 readdir)。
生产路径 ≈ 65ms(stat) + <20ms(索引读) + 40–50ms(rdprog×595) + <20ms(build_rows)
≈ 0.13–0.15s(对比 3.0s)。

索引格式 v1(手工小端序列化,不依赖结构体布局):
header(magic/version/header_size/root_len) + root + dir 记录×(rel,mtime_s,
mtime_ns,books_here) + book 记录×(rel,size,fmt) + trailer(ndirs,nbooks,flags,
depth_skipped,crc32,magic)。truncated/depth_skipped 存在 flags/字段里,载入时
如实重建提示条;**不存阅读进度 ts**,载入后每本由 rdprog_load 现读再排序。

失效语义(存每个扫描过的目录,不只 level-1):
- level-2 新增 / level-3 深处新增 / 删书 → 该目录 mtime 变 → 重扫;
- 一级目录改名 → 父目录 mtime 变 + 旧目录 stat ENOENT → 重扫;
- 超深(>6 层)未扫描子树不在索引内,其内部变化不自动失效(本来就不显示),
  点「刷新」更新 depth_skipped 计数;mtime 精度受限(exFAT 10ms/FAT32 2s)的
  同刻度变更也可能漏判,点「刷新」必可纠正。
M29 语义不变:首次进书架=索引或全扫;回书架=不扫;刷新=删索引+全扫(之后原子
写回新索引)。

安全/健壮(宁可慢不可错):
- 写:shelf.idx.tmp 全量写完(尽力 fsync)后 rename;中途被杀只剩 tmp,正式索引
  要么旧要么完整新;rename 覆盖失败先删旧再改名;fflush/写错即放弃,绝不影响书架。
- 读:O_NOFOLLOW|O_NONBLOCK + S_ISREG(FIFO/符号链接直接拒绝不阻塞);CRC32 覆盖
  全部记录;rel 做绝对路径/“..”/内嵌 NUL/长度校验;版本/根路径/计数/逐目录 mtime
  任一不符即重扫;载入崩溃路径全部 goto 清理,不改书架可用性。
- writer_open 失败(卡满/写保护/无 MeowData)→ 扫描不采集目录、照常出书;
  commit 失败(含只读目录 rename 失败)不留正式索引。

测试:test/shelf_index_smoke 154 pass(含四种变更场景失效、截断/字节翻转/垃圾/
缺失/FIFO/符号链接/版本不符/根路径不符、ts 现读、truncated/depth_skipped 重建、
只读目录写失败、空 Books、root 缺失、.tmp 残渣),ASan+UBSan 干净;
test/shelf_index_bench 模拟 434 目录/595 本/51 rar:scan/load 均 595 本、
truncated=0/depth_skipped=0、索引 72,156B。build_reader.sh 全绿 0 warning
(core 6846 / seg 19 / chapter 71 / shelf_group 63 / index 154 / font 14 /
loaders PASS)。shelf_bench_ondev 增加 index_load/scan_save 计时与
MEOW_BENCH_IDX=miss(真机首开命中 vs 未命中对比,待有卡复核)。

交付:app.json 0.1.4/code5 → 0.1.5/code6;reader md5 df4cadcf
1,711,924B;mpk md5 fec7b1bb 814,535B;未签名不推卡。
This commit is contained in:
AskaEth committed 2026-11-10 18:14:52 +08:00
1 parent 3f09f7c137
commit e0dfab95c3
11 files changed
+1388 -29

No files matched your search

@@ -15,13 +15,44 @@
*/
#include "reader_internal.h"
#include "app/shelf_model.h"
#include "app/shelf_index.h"
#include "meow_ui_internal.h"
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/stat.h>
static uint64_t now_ms(void) { return meow_ui_time_ms(); }
/* M32:索引命中/未命中的耗时(不含 UI)。先在本次运行开头测:
* - index_load:读到上次运行写下的索引(首次运行 miss);
* - scan_save :全量扫描 + 落盘(生产「无索引首次进入」)。
* 之后 shelf_on_enter 会走生产路径:有索引=命中(默认),MEOW_BENCH_IDX=miss 时先删索引。 */
static void bench_index(void)
{
static shelf_book books[SHELF_MAX_BOOKS];
shelf_scan_report rep;
uint64_t t0 = now_ms();
int hit = shelf_index_load(READER_SD_ROOT "/Books", books, SHELF_MAX_BOOKS, NULL, &rep);
uint64_t t1 = now_ms();
printf("BENCH index_load_ms %llu (hit=%d n=%d trunc=%d skip=%d)\n",
(unsigned long long)(t1 - t0), hit, rep.nbooks, rep.truncated, rep.depth_skipped);
t0 = now_ms();
shelf_idx_writer *w = shelf_index_writer_open(READER_SD_ROOT "/Books");
int n = shelf_scan_books_ex(READER_SD_ROOT "/Books", books, SHELF_MAX_BOOKS, NULL, &rep,
w ? shelf_index_writer_dir : NULL, w);
int ok = w ? shelf_index_writer_commit(w, books, n, &rep) : 0;
t1 = now_ms();
char ip[500];
struct stat st;
long long isz = (shelf_index_path(READER_SD_ROOT "/Books", ip, sizeof ip) == 0 &&
stat(ip, &st) == 0) ? (long long)st.st_size : -1;
printf("BENCH scan_save_ms %llu (n=%d commit=%d index_bytes=%lld)\n",
(unsigned long long)(t1 - t0), n, ok, isz);
fflush(stdout);
}
static void bench_scan(void)
{
static shelf_book books[SHELF_MAX_BOOKS];
@@ -40,8 +71,17 @@ int main(int argc, char **argv)
if (meow_ui_init(&o) != 0) { printf("BENCH init_fail\n"); return 1; }
meow_theme_add_listener(NULL);
bench_index(); /* M32:索引命中/未命中 + 扫描落盘(不含 UI) */
bench_scan(); /* 纯扫描基线(不建 UI) */
/* 首次进书架:默认走索引命中;MEOW_BENCH_IDX=miss 先删索引 → 走全量扫描(对比用)。 */
const char *eidx = getenv("MEOW_BENCH_IDX");
if (eidx && strcmp(eidx, "miss") == 0) {
shelf_index_remove(READER_SD_ROOT "/Books");
printf("BENCH index_removed (miss path)\n");
fflush(stdout);
}
shelf_view_create();
reader_view_create();
settings_view_create();
@@ -0,0 +1,254 @@
/* test/shelf_index_bench.c — M32 索引方案可行性计时(宿主,纯 C + clock_gettime)。
*
* 用法:
* ./shelf_index_bench_host [bench|indexonly] [root]
* bench —— 建虚拟书库(默认 452 目录 / 688 本,贴合真机 434/595 量级)并计时
* indexonly —— 只载入已有索引一次(供 `strace -e getdents64` 断言「载入不 readdir」)
*
* 为什么宿主数字不能直接当真机用:宿主是 ext4/f2fs 热缓存(每次 stat ~1µs),真机 TF 卡是
* exFAT(M28 实测 1079 次 stat = 160–170ms,约 150µs/次)。本工具用来量**调用次数与代码路径**
* 的相对成本;绝对毫秒按 M28 的每次 stat/条目成本外推(见报文)。
*
* 计时口径:
* - stat_all :先 opendir 走一遍把目录列表收好(不计时),再单独循环 stat(计时的只有 stat);
* - full_scan :shelf_scan_books_ex(生产全量扫描形态,ts_fn=NULL);
* - scan+save :全量扫描 + 索引落盘(生产「无索引首次进入」路径);
* - index_load :shelf_index_load 命中(生产「有索引首次进入」路径,ts_fn=NULL 以隔离进度读);
* - index_read :单次 open+fstat+read 整个索引文件(校验之前的纯 IO 成本)。
*/
#include "shelf_model.h"
#include "shelf_index.h"
#include <dirent.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/stat.h>
#include <time.h>
#include <unistd.h>
static char g_cfgdir[500];
const char *meow_config_dir(void) { return g_cfgdir; }
static double now_ms(void)
{
struct timespec ts;
clock_gettime(CLOCK_MONOTONIC, &ts);
return (double)ts.tv_sec * 1000.0 + (double)ts.tv_nsec / 1e6;
}
static void mkdirs(const char *path)
{
char t[600];
snprintf(t, sizeof t, "%s", path);
for (char *p = t + 1; *p; p++) {
if (*p == '/') { *p = 0; mkdir(t, 0755); *p = '/'; }
}
mkdir(t, 0755);
}
static void put_file(const char *path, long size)
{
FILE *f = fopen(path, "wb");
if (!f) return;
fseek(f, size > 0 ? size - 1 : 0, SEEK_SET);
fputc('x', f);
fclose(f);
}
/* 虚拟书库:精确复刻真机量级 —— 434 个目录 / 595 本书 / 51 个 .rar(不支持格式)。
* 根 1 + 一级 10 + 「400+本高质量完本合集」下 40 分类 + 382 作品目录 + 轻小说/子目录 1 = 434;
* 书 = 382 作品 × 1 + 其它 9 个一级目录 × 20 + 根 3 + 子目录 30 = 595。 */
static int make_library(const char *root)
{
char p[600];
DIR *d = opendir(root);
if (d) { closedir(d); return 0; } /* 已存在:复用 */
mkdirs(root);
const char *l1[] = { "400+本高质量完本合集", "轻小说", "玄幻", "都市", "科幻",
"历史", "武侠", "游戏", "悬疑", "同人科幻" };
for (size_t i = 0; i < sizeof l1 / sizeof l1[0]; i++) {
snprintf(p, sizeof p, "%s/%s", root, l1[i]); mkdirs(p);
}
int works = 0, rar = 0;
/* 40 个分类目录,382 个作品目录(前 22 个分类各 10 个,其余各 9 个):
* 1 根 + 10 一级 + 40 分类 + 382 作品 + 1 子目录 = 434 个目录。 */
for (int c = 1; c <= 40; c++) {
snprintf(p, sizeof p, "%s/400+本高质量完本合集/%02d、分类%d", root, c, c); mkdirs(p);
int nw = (c <= 22) ? 10 : 9;
for (int w = 1; w <= nw; w++, works++) {
char wd[600];
snprintf(wd, sizeof wd, "%s/400+本高质量完本合集/%02d、分类%d/%02d、作品%d",
root, c, c, w, w);
mkdirs(wd);
snprintf(p, sizeof p, "%s/第1卷.txt", wd);
put_file(p, 1024 + (long)((works * 7919) % 200000));
if ((works % 7) == 0 && rar < 51) { snprintf(p, sizeof p, "%s/合集.rar", wd); put_file(p, 1000); rar++; }
}
}
for (size_t i = 1; i < sizeof l1 / sizeof l1[0]; i++) {
for (int b = 1; b <= 20; b++) {
snprintf(p, sizeof p, "%s/%s/书%d.txt", root, l1[i], b);
put_file(p, 4096);
}
}
for (int b = 1; b <= 3; b++) {
snprintf(p, sizeof p, "%s/根下%d.txt", root, b);
put_file(p, 2048);
}
snprintf(p, sizeof p, "%s/轻小说/子目录", root); mkdirs(p);
for (int b = 1; b <= 30; b++) {
snprintf(p, sizeof p, "%s/轻小说/子目录/深%d.txt", root, b);
put_file(p, 4096);
}
/* 补齐 51 个 rar(若作品目录里不够,放到一级目录) */
while (rar < 51) {
snprintf(p, sizeof p, "%s/玄幻/压缩%d.rar", root, rar);
put_file(p, 100);
rar++;
}
return 1;
}
/* 收集目录列表(不计时;用于单独量 stat 成本)。 */
static int collect_dirs(const char *dir, char paths[][512], int max, int n, int depth)
{
if (depth > 8 || n >= max) return n;
DIR *d = opendir(dir);
if (!d) return n;
struct dirent *e;
while ((e = readdir(d))) {
if (e->d_name[0] == '.') continue;
char path[512];
if (snprintf(path, sizeof path, "%s/%s", dir, e->d_name) >= (int)sizeof path) continue;
struct stat st;
if (stat(path, &st) != 0) continue;
if (S_ISDIR(st.st_mode) && n < max) {
snprintf(paths[n], 512, "%s", path);
n = collect_dirs(path, paths, max, n + 1, depth + 1);
}
}
closedir(d);
return n;
}
static int count_files(const char *dir, int depth)
{
if (depth > 8) return 0;
DIR *d = opendir(dir);
if (!d) return 0;
int n = 0;
struct dirent *e;
while ((e = readdir(d))) {
if (e->d_name[0] == '.') continue;
char path[512];
snprintf(path, sizeof path, "%s/%s", dir, e->d_name);
struct stat st;
if (stat(path, &st) != 0) continue;
if (S_ISDIR(st.st_mode)) n += count_files(path, depth + 1);
else n++;
}
closedir(d);
return n;
}
int main(int argc, char **argv)
{
const char *mode = (argc > 1) ? argv[1] : "bench";
const char *tmp = getenv("TMPDIR");
if (!tmp || !tmp[0]) tmp = "/tmp";
char base[500], root[520];
if (argc > 2) snprintf(base, sizeof base, "%s", argv[2]);
else snprintf(base, sizeof base, "%s/m32bench", tmp);
snprintf(root, sizeof root, "%s/Books", base);
snprintf(g_cfgdir, sizeof g_cfgdir, "%s/data", base);
mkdirs(g_cfgdir);
static shelf_book books[SHELF_MAX_BOOKS];
shelf_scan_report rep;
if (strcmp(mode, "indexonly") == 0) {
int hit = shelf_index_load(root, books, SHELF_MAX_BOOKS, NULL, &rep);
printf("indexonly hit=%d nbooks=%d\n", hit, rep.nbooks);
return hit ? 0 : 1;
}
int created = make_library(root);
static char dirs[4096][512];
snprintf(dirs[0], 512, "%s", root);
int nd = 1 + collect_dirs(root, dirs, 4096, 0, 0);
int nfiles = count_files(root, 0);
printf("library: %s (%s) dirs=%d files=%d\n", root, created ? "created" : "reused", nd, nfiles);
/* ① 目录逐个 stat(校验索引的唯一 IO) */
struct stat st;
int reps = 20;
double t0 = now_ms();
for (int r = 0; r < reps; r++)
for (int i = 0; i < nd; i++) stat(dirs[i], &st);
double t1 = now_ms();
printf("stat_all_dirs: %d x%d = %.2f ms total, %.3f ms/pass, %.2f us/stat\n",
nd, reps, t1 - t0, (t1 - t0) / reps, (t1 - t0) * 1000.0 / (reps * nd));
/* ② 全量扫描(无 ts) */
double best_scan = 1e9, best_save = 1e9;
int n = 0;
for (int r = 0; r < 3; r++) {
double a = now_ms();
n = shelf_scan_books_ex(root, books, SHELF_MAX_BOOKS, NULL, &rep, NULL, NULL);
double b = now_ms();
if (b - a < best_scan) best_scan = b - a;
}
printf("full_scan: books=%d truncated=%d depth_skipped=%d best=%.2f ms\n",
n, rep.truncated, rep.depth_skipped, best_scan);
int scan_trunc = rep.truncated, scan_skip = rep.depth_skipped;
/* ③ 全量扫描 + 写索引 */
for (int r = 0; r < 3; r++) {
double a = now_ms();
shelf_idx_writer *w = shelf_index_writer_open(root);
int n2 = shelf_scan_books_ex(root, books, SHELF_MAX_BOOKS, NULL, &rep,
w ? shelf_index_writer_dir : NULL, w);
int ok = w ? shelf_index_writer_commit(w, books, n2, &rep) : 0;
double b = now_ms();
if (b - a < best_save) best_save = b - a;
if (r == 0) printf("scan+save: scan_books=%d index_commit=%d\n", n2, ok);
}
printf("scan+save: best=%.2f ms\n", best_save);
/* ④ 索引命中(不含 ts 读取) */
double best_load = 1e9;
int hit = 0, ln = 0;
for (int r = 0; r < 5; r++) {
double a = now_ms();
hit = shelf_index_load(root, books, SHELF_MAX_BOOKS, NULL, &rep);
double b = now_ms();
ln = rep.nbooks;
if (b - a < best_load) best_load = b - a;
}
printf("index_load: hit=%d nbooks=%d best=%.2f ms\n", hit, ln, best_load);
/* ⑤ 索引文件大小 + 单次顺序读成本 */
char ip[600];
if (shelf_index_path(root, ip, sizeof ip) == 0 && stat(ip, &st) == 0) {
char *buf = (char *)malloc((size_t)st.st_size);
double a = now_ms();
for (int r = 0; r < 20 && buf; r++) {
FILE *f = fopen(ip, "rb");
if (f) { if (fread(buf, 1, (size_t)st.st_size, f) != (size_t)st.st_size) { /* 短读 */ } fclose(f); }
}
double b = now_ms();
printf("index_file: %lld bytes, open+read %.3f ms/pass\n",
(long long)st.st_size, buf ? (b - a) / 20 : 0.0);
free(buf);
}
/* ⑥ 生产路径估算:索引命中 + 595 次 rdprog_load(M28 实测 40–50ms,未在此重复测) */
printf("NOTE 生产路径 = index_load + rdprog_load×nbooks(M28: 40–50ms)+ build_rows(<20ms)\n");
/* ⑦ 模拟真机书库的正确性断言:595 本全可见、无截断、无超深跳过,索引与扫描同数。 */
int pass = (n == 595 && ln == 595 && hit == 1 && scan_trunc == 0 && scan_skip == 0);
printf("CHECK library_sim scan_books=%d load_books=%d hit=%d truncated=%d depth_skipped=%d -> %s\n",
n, ln, hit, scan_trunc, scan_skip, pass ? "PASS" : "FAIL");
return pass ? 0 : 1;
}
@@ -0,0 +1,426 @@
/* test/shelf_index_smoke.c — M32 书架索引宿主单测(纯 C,无设备/SDK 依赖)。
*
* 覆盖:
* I1 全量扫描 → 写索引 → 载入:书数/书名/分组/path/fmt/size/ts 全部与扫描一致
* I2 载入不缓存阅读进度:同一索引换 ts_fn 再载入,ts 立刻反映新值(含真实 rdprog 文件)
* I3 四种书库变更都必须失效(fail-open 重扫):
* level-2 新增 / level-3 深处新增 / level-1 目录改名 / 删书
* I4 坏档 fail-open:截断 / 中间字节翻转 / 垃圾内容 / 文件缺失 / 索引路径是目录 /
* 版本不符 / Books 根路径不符 —— 一律返回 0,不崩
* I5 truncated / depth_skipped 如实重建(索引里存了这两个值)
* I6 写盘失败不影响书架:索引目录不可写时 writer_open 返回 NULL、扫描照常出书;
* commit 失败不留下半截索引
* I7 空 Books / Books 不存在:空索引可载入;root_missing 不写索引
* I8 载入路径不做 readdir(由 build 验证步骤用 strace 另行断言)
*/
#include "shelf_model.h"
#include "shelf_index.h"
#include "progress.h"
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/stat.h>
#include <sys/types.h>
#include <unistd.h>
static int g_pass = 0, g_fail = 0;
#define CHECK(cond, msg) do { \
if (cond) g_pass++; \
else { g_fail++; printf("FAIL %s:%d %s\n", __FILE__, __LINE__, msg); } \
} while (0)
static char g_cfgdir[400];
const char *meow_config_dir(void) { return g_cfgdir; }
static void mkdirs(const char *path)
{
char t[400];
snprintf(t, sizeof t, "%s", path);
for (char *p = t + 1; *p; p++) {
if (*p == '/') { *p = 0; mkdir(t, 0755); *p = '/'; }
}
mkdir(t, 0755);
}
static void write_file(const char *path, const char *body)
{
FILE *f = fopen(path, "wb");
if (!f) { printf("无法创建 %s\n", path); exit(2); }
fputs(body, f);
fclose(f);
}
/* 进度按文件名给(I2:换 ts_fn 必须立即生效,证明索引没缓存 ts)。 */
static uint64_t g_ts_boost;
static uint64_t ts_of(const char *path)
{
const char *b = strrchr(path, '/');
b = b ? b + 1 : path;
if (strcmp(b, "1.txt") == 0) return 100 + g_ts_boost;
if (strcmp(b, "2.epub") == 0) return 50;
return 0;
}
/* 真实进度文件口径(I2 第二段)。 */
static uint64_t prog_ts_of(const char *path)
{
rdprog p;
return (rdprog_load(path, &p) == 1) ? p.ts : 0;
}
static const shelf_book *find_path(const shelf_book *books, int n, const char *path)
{
for (int i = 0; i < n; i++) if (strcmp(books[i].path, path) == 0) return &books[i];
return NULL;
}
static const shelf_book *find_title(const shelf_book *books, int n, const char *title)
{
for (int i = 0; i < n; i++) if (strcmp(books[i].title, title) == 0) return &books[i];
return NULL;
}
static int idx_path(const char *root, char *out, int outsz)
{
return shelf_index_path(root, out, outsz);
}
static int scan_and_save(const char *root, shelf_book *books, int cap, shelf_scan_report *rep)
{
shelf_idx_writer *w = shelf_index_writer_open(root);
CHECK(w != NULL, "writer_open 成功");
int n = shelf_scan_books_ex(root, books, cap, ts_of, rep,
w ? shelf_index_writer_dir : NULL, w);
if (w) CHECK(shelf_index_writer_commit(w, books, n, rep) == 1, "writer_commit 成功");
return n;
}
/* 建一棵贴合真机布局的树(含 level-3 深处、超深目录、不支持格式、空目录)。 */
static void make_tree(const char *root)
{
char p[400];
mkdirs(root);
snprintf(p, sizeof p, "%s/轻小说", root); mkdirs(p);
snprintf(p, sizeof p, "%s/玄幻", root); mkdirs(p);
snprintf(p, sizeof p, "%s/B/sub", root); mkdirs(p);
snprintf(p, sizeof p, "%s/D/E/F", root); mkdirs(p);
snprintf(p, sizeof p, "%s/G/H/I/J/K/L/M", root); mkdirs(p);
snprintf(p, sizeof p, "%s/400+本高质量完本合集/01、玄幻/01、斗破", root); mkdirs(p);
snprintf(p, sizeof p, "%s/空目录", root); mkdirs(p);
snprintf(p, sizeof p, "%s/轻小说/1.txt", root); write_file(p, "a");
snprintf(p, sizeof p, "%s/轻小说/2.epub", root); write_file(p, "bb");
snprintf(p, sizeof p, "%s/轻小说/notes.md", root); write_file(p, "not a book");
snprintf(p, sizeof p, "%s/玄幻/3.txt", root); write_file(p, "ccc");
snprintf(p, sizeof p, "%s/根下.txt", root); write_file(p, "root book");
snprintf(p, sizeof p, "%s/400+本高质量完本合集/01、玄幻/01、斗破/《斗破》.txt", root);
write_file(p, "doupo");
snprintf(p, sizeof p, "%s/B/sub/5.mobi", root); write_file(p, "mobi");
snprintf(p, sizeof p, "%s/D/E/F/6.txt", root); write_file(p, "deep3");
snprintf(p, sizeof p, "%s/G/H/I/J/K/L/M/7.txt", root); write_file(p, "too deep");
}
static void test_roundtrip(const char *root)
{
static shelf_book scanb[SHELF_MAX_BOOKS], loadb[SHELF_MAX_BOOKS];
shelf_scan_report r1, r2;
int n = scan_and_save(root, scanb, SHELF_MAX_BOOKS, &r1);
CHECK(n == 7, "扫描到 7 本(7.txt 超深不计)");
CHECK(r1.depth_skipped == 1 && r1.truncated == 0, "深度超限计数=1");
char ip[500];
CHECK(idx_path(root, ip, sizeof ip) == 0, "索引路径可得");
struct stat st;
CHECK(stat(ip, &st) == 0 && st.st_size > 0, "索引文件已落盘");
printf("index: %s (%lld bytes)\n", ip, (long long)st.st_size);
int n2 = shelf_index_load(root, loadb, SHELF_MAX_BOOKS, ts_of, &r2);
CHECK(n2 == 1, "索引命中");
CHECK(r2.nbooks == n && r2.truncated == r1.truncated &&
r2.depth_skipped == r1.depth_skipped && r2.root_missing == 0, "报告重建一致");
for (int i = 0; i < n; i++) {
const shelf_book *b = find_path(loadb, r2.nbooks, scanb[i].path);
CHECK(b != NULL, "载入含同一 path");
if (!b) continue;
CHECK(strcmp(b->title, scanb[i].title) == 0, "title 一致");
CHECK(strcmp(b->group, scanb[i].group) == 0, "group 一致");
CHECK(strcmp(b->top, scanb[i].top) == 0, "top 一致");
CHECK(strcmp(b->sub, scanb[i].sub) == 0, "sub 一致");
CHECK(strcmp(b->sub_folder, scanb[i].sub_folder) == 0, "sub_folder 一致");
CHECK(b->fmt == scanb[i].fmt && b->size == scanb[i].size, "fmt/size 一致");
CHECK(b->ts == scanb[i].ts, "ts 一致");
}
/* 排序:最近阅读优先 */
CHECK(r2.nbooks >= 2 && strcmp(loadb[0].title, "1") == 0,
"载入后按 ts 排序(1.txt 100 在最前)");
}
/* I2:索引不缓存 ts —— 同一份索引,换 ts_fn 载入必须立即反映新值。 */
static void test_ts_not_cached(const char *root)
{
static shelf_book b[SHELF_MAX_BOOKS];
shelf_scan_report rep;
g_ts_boost = 0;
CHECK(scan_and_save(root, b, SHELF_MAX_BOOKS, &rep) == 7, "重新扫描写索引");
g_ts_boost = 900;
static shelf_book lb[SHELF_MAX_BOOKS];
CHECK(shelf_index_load(root, lb, SHELF_MAX_BOOKS, ts_of, &rep) == 1, "索引命中");
const shelf_book *x = find_title(lb, rep.nbooks, "1");
CHECK(x && x->ts == 1000, "载入后 ts 用当前 ts_fn(100+900),不是索引里的旧值");
CHECK(x && strstr(x->sub, "读") != NULL, "副题随新 ts 重建(· 读)");
g_ts_boost = 0;
/* 真实进度文件:写 rdprog 后同一索引载入,ts/副题立即更新。 */
char path[400];
snprintf(path, sizeof path, "%s/根下.txt", root);
rdprog p;
memset(&p, 0, sizeof p);
p.valid = 1; p.seg = 2; p.ch = 9; p.fs = 20; p.ls = 130; p.enc = -1; p.ts = 777777;
CHECK(rdprog_save(path, &p) == 0, "写入测试进度");
CHECK(shelf_index_load(root, lb, SHELF_MAX_BOOKS, prog_ts_of, &rep) == 1, "索引命中(真实进度)");
const shelf_book *y = find_title(lb, rep.nbooks, "根下");
CHECK(y && y->ts == 777777, "ts 来自 rdprog_load 现读");
/* 删掉进度文件(注意 rdprog_save 的参数是书路径,进度文件另按 hash 落盘)。 */
char pdir[400], phash[24], ppath[440];
if (rdprog_dir(pdir, sizeof pdir) == 0) {
rdprog_hash(path, phash, sizeof phash);
snprintf(ppath, sizeof ppath, "%s/%s", pdir, phash);
unlink(ppath);
}
}
/* I3:四种变更都必须让索引失效。每次变更后:先断言 load=0(不静默用陈旧索引),
* 再全量重扫 + 写新索引,断言变化可见。 */
static void test_invalidation(const char *root)
{
static shelf_book b[SHELF_MAX_BOOKS];
shelf_scan_report rep;
char p[400], q[400];
CHECK(scan_and_save(root, b, SHELF_MAX_BOOKS, &rep) > 0, "基线扫描写索引");
/* ① level-2 直接新增 */
usleep(20000);
snprintf(p, sizeof p, "%s/轻小说/新书.txt", root); write_file(p, "new");
CHECK(shelf_index_load(root, b, SHELF_MAX_BOOKS, ts_of, &rep) == 0,
"level-2 新增 → 索引失效");
int n = scan_and_save(root, b, SHELF_MAX_BOOKS, &rep);
CHECK(find_title(b, n, "新书") != NULL, "重扫后新书可见");
/* ② level-3 深处新增(Books/400+.../01、玄幻/01、斗破/) */
usleep(20000);
snprintf(p, sizeof p, "%s/400+本高质量完本合集/01、玄幻/01、斗破/新书2.txt", root);
write_file(p, "new2");
CHECK(shelf_index_load(root, b, SHELF_MAX_BOOKS, ts_of, &rep) == 0,
"level-3 深处新增 → 索引失效");
n = scan_and_save(root, b, SHELF_MAX_BOOKS, &rep);
CHECK(find_title(b, n, "新书2") != NULL, "重扫后深处新书可见");
/* ③ 一级目录改名 */
usleep(20000);
snprintf(p, sizeof p, "%s/玄幻", root);
snprintf(q, sizeof q, "%s/玄幻改", root);
CHECK(rename(p, q) == 0, "改名一级目录");
CHECK(shelf_index_load(root, b, SHELF_MAX_BOOKS, ts_of, &rep) == 0,
"一级目录改名 → 索引失效");
n = scan_and_save(root, b, SHELF_MAX_BOOKS, &rep);
const shelf_book *m = find_title(b, n, "3");
CHECK(m && strcmp(m->top, "玄幻改") == 0 && strcmp(m->group, "玄幻改") == 0,
"重扫后分组键跟随新目录名");
/* ④ 删书 */
usleep(20000);
snprintf(p, sizeof p, "%s/根下.txt", root);
CHECK(unlink(p) == 0, "删除一本书");
CHECK(shelf_index_load(root, b, SHELF_MAX_BOOKS, ts_of, &rep) == 0,
"删书 → 索引失效");
n = scan_and_save(root, b, SHELF_MAX_BOOKS, &rep);
CHECK(find_title(b, n, "根下") == NULL, "重扫后删掉的书消失");
}
static void test_bad_index(const char *root)
{
static shelf_book b[SHELF_MAX_BOOKS];
shelf_scan_report rep;
char ip[500];
CHECK(idx_path(root, ip, sizeof ip) == 0, "索引路径可得");
CHECK(scan_and_save(root, b, SHELF_MAX_BOOKS, &rep) > 0, "基线扫描写索引");
/* 截断:去掉 tail 10 字节(trailer 没了) */
struct stat st;
CHECK(stat(ip, &st) == 0, "索引存在");
CHECK(truncate(ip, st.st_size - 10) == 0, "截断索引");
CHECK(shelf_index_load(root, b, SHELF_MAX_BOOKS, ts_of, &rep) == 0, "截断 → 拒绝");
/* 中间字节翻转:CRC 必须抓到 */
CHECK(scan_and_save(root, b, SHELF_MAX_BOOKS, &rep) > 0, "重写索引");
FILE *f = fopen(ip, "r+b");
CHECK(f != NULL, "打开索引改写");
if (f) {
CHECK(fseek(f, st.st_size / 2, SEEK_SET) == 0, "定位中部");
int c = fgetc(f);
CHECK(c != EOF, "读中部字节");
CHECK(fseek(f, st.st_size / 2, SEEK_SET) == 0, "回定位");
fputc(c ^ 0x01, f);
fclose(f);
}
CHECK(shelf_index_load(root, b, SHELF_MAX_BOOKS, ts_of, &rep) == 0, "字节翻转 → 拒绝");
/* 垃圾内容 */
write_file(ip, "this is not a shelf index at all, just garbage garbage garbage");
CHECK(shelf_index_load(root, b, SHELF_MAX_BOOKS, ts_of, &rep) == 0, "垃圾内容 → 拒绝");
/* 文件缺失 */
unlink(ip);
CHECK(shelf_index_load(root, b, SHELF_MAX_BOOKS, ts_of, &rep) == 0, "文件缺失 → 拒绝");
/* FIFO:O_NONBLOCK 必须直接拒绝(不能挂死在 open 上) */
CHECK(mkfifo(ip, 0644) == 0, "建 FIFO 假索引");
CHECK(shelf_index_load(root, b, SHELF_MAX_BOOKS, ts_of, &rep) == 0, "FIFO → 拒绝且不阻塞");
unlink(ip);
/* 符号链接:O_NOFOLLOW 只认普通文件,不跟随 */
CHECK(symlink("/etc/hostname", ip) == 0, "建符号链接假索引");
CHECK(shelf_index_load(root, b, SHELF_MAX_BOOKS, ts_of, &rep) == 0, "符号链接 → 拒绝");
unlink(ip);
/* 索引路径是目录 */
mkdir(ip, 0755);
CHECK(shelf_index_load(root, b, SHELF_MAX_BOOKS, ts_of, &rep) == 0, "索引是目录 → 拒绝");
rmdir(ip);
/* 扫描中途被杀留下的 .tmp 残渣:绝不能影响正式索引的载入 */
CHECK(scan_and_save(root, b, SHELF_MAX_BOOKS, &rep) > 0, "重写索引");
char tmpip[520];
snprintf(tmpip, sizeof tmpip, "%s.tmp", ip);
write_file(tmpip, "half written index leftover");
CHECK(shelf_index_load(root, b, SHELF_MAX_BOOKS, ts_of, &rep) == 1, ".tmp 残渣不影响载入");
unlink(tmpip);
/* 版本不符:先把版本字段改成 2 */
CHECK(scan_and_save(root, b, SHELF_MAX_BOOKS, &rep) > 0, "重写索引");
f = fopen(ip, "r+b");
CHECK(f != NULL, "打开索引改版本");
if (f) {
unsigned char v[4] = { 2, 0, 0, 0 };
CHECK(fseek(f, 8, SEEK_SET) == 0, "定位版本字段");
fwrite(v, 1, sizeof v, f);
fclose(f);
}
CHECK(shelf_index_load(root, b, SHELF_MAX_BOOKS, ts_of, &rep) == 0, "版本不符 → 拒绝");
/* Books 根路径不符(同长度不同路径) */
CHECK(scan_and_save(root, b, SHELF_MAX_BOOKS, &rep) > 0, "重写索引");
char root2[400];
snprintf(root2, sizeof root2, "%s", root);
root2[strlen(root2) - 1] = 'z'; /* Books → Bookz */
CHECK(shelf_index_load(root2, b, SHELF_MAX_BOOKS, ts_of, &rep) == 0, "根路径不符 → 拒绝");
}
/* I5:cap 截断 + 超深计数都要从索引如实重建。 */
static void test_report_rebuild(const char *root)
{
static shelf_book b[SHELF_MAX_BOOKS];
shelf_scan_report r1;
int n = scan_and_save(root, b, 3, &r1);
CHECK(n == 3 && r1.truncated == 1, "cap=3 扫描截断标记");
CHECK(shelf_index_load(root, b, 3, ts_of, &r1) == 1, "截断索引仍可命中");
CHECK(r1.truncated == 1 && r1.nbooks == 3, "截断标记如实重建");
}
/* I6:索引不可写 → 扫描照常;commit 失败无残留。 */
static void test_write_failure(const char *root)
{
static shelf_book b[SHELF_MAX_BOOKS];
shelf_scan_report rep;
char ip[500];
CHECK(idx_path(root, ip, sizeof ip) == 0, "索引路径可得");
char idir[500];
snprintf(idir, sizeof idir, "%s", ip);
char *slash = strrchr(idir, '/');
if (slash) *slash = 0;
mkdirs(idir);
if (geteuid() == 0) { printf("(skip 写失败用例:root 不受权限位限制)\n"); return; }
CHECK(chmod(idir, 0500) == 0, "索引目录改为只读");
shelf_idx_writer *w = shelf_index_writer_open(root);
CHECK(w == NULL, "只读目录 → writer_open 返回 NULL(不崩)");
/* 生产路径:w=NULL → 扫描不采集目录,仍必须出书。 */
int n = shelf_scan_books_ex(root, b, SHELF_MAX_BOOKS, ts_of, &rep, NULL, NULL);
CHECK(n > 0 && rep.root_missing == 0, "索引不可写时书架扫描照常");
CHECK(chmod(idir, 0755) == 0, "恢复目录权限");
/* commit 失败:打开 writer(tmp 已建)后再把目录改只读 → rename 失败 → 返回 0 且无残留。 */
unlink(ip); /* 先清掉前面用例留下的正式索引,才能断言「失败后不存在」 */
w = shelf_index_writer_open(root);
CHECK(w != NULL, "可写时 writer_open 成功");
if (w) {
CHECK(chmod(idir, 0500) == 0, "提交前目录改只读");
int navail = shelf_scan_books_ex(root, b, SHELF_MAX_BOOKS, ts_of, &rep,
shelf_index_writer_dir, w);
int ok = shelf_index_writer_commit(w, b, navail, &rep);
CHECK(ok == 0, "rename 失败 → commit 返回 0");
CHECK(chmod(idir, 0755) == 0, "恢复目录权限");
struct stat st;
CHECK(stat(ip, &st) != 0, "commit 失败不留正式索引");
}
}
/* I7:空 Books 可缓存;root 不存在不写索引。 */
static void test_empty_and_missing(const char *root)
{
static shelf_book b[SHELF_MAX_BOOKS];
shelf_scan_report rep;
char empty[400];
snprintf(empty, sizeof empty, "%s_empty/Books", root);
mkdirs(empty);
int n = scan_and_save(empty, b, SHELF_MAX_BOOKS, &rep);
CHECK(n == 0 && rep.root_missing == 0, "空 Books 扫描 0 本");
CHECK(shelf_index_load(empty, b, SHELF_MAX_BOOKS, ts_of, &rep) == 1, "空索引可命中");
CHECK(rep.nbooks == 0, "空索引 0 本");
char missing[400];
snprintf(missing, sizeof missing, "%s_nope/Books", root);
CHECK(shelf_index_load(missing, b, SHELF_MAX_BOOKS, ts_of, &rep) == 0, "根不存在 → 拒绝");
/* writer_open 只依赖 App 数据目录(不 stat Books):根不存在时也能打开,
* 但扫描不到任何目录(ndirs=0)→ commit 必须失败且不发布半截索引。 */
shelf_scan_report rm;
shelf_idx_writer *w = shelf_index_writer_open(missing);
int mn = shelf_scan_books_ex(missing, b, SHELF_MAX_BOOKS, ts_of, &rm,
w ? shelf_index_writer_dir : NULL, w);
CHECK(mn == 0 && rm.root_missing == 1, "根不存在 → root_missing");
if (w) CHECK(shelf_index_writer_commit(w, b, mn, &rm) == 0, "根不存在 → commit 拒绝空索引");
}
int main(void)
{
const char *base = getenv("TMPDIR");
if (!base || !base[0]) base = "/tmp";
char tmpl[400];
snprintf(tmpl, sizeof tmpl, "%s/m32_index_XXXXXX", base);
char *tmp = mkdtemp(tmpl);
if (!tmp) { printf("mkdtemp 失败\n"); return 2; }
snprintf(g_cfgdir, sizeof g_cfgdir, "%s/data", tmp); /* 索引落 <tmp>/index */
mkdirs(g_cfgdir);
char root[400];
snprintf(root, sizeof root, "%s/Books", tmp);
make_tree(root);
test_roundtrip(root);
test_ts_not_cached(root);
test_invalidation(root);
test_bad_index(root);
test_report_rebuild(root);
test_write_failure(root);
test_empty_and_missing(root);
printf("shelf_index_smoke: %d pass, %d fail\n", g_pass, g_fail);
char cmd[512];
snprintf(cmd, sizeof cmd, "rm -rf '%s' '%s_empty' '%s_nope'", tmp, root, root);
if (system(cmd) != 0) { /* 忽略清理失败 */ }
return g_fail ? 1 : 0;
}