Update tesseract datapaths

Update USAGE.md

Dockerfile.arm64

@@ -7,7 +7,7 @@ RUN cmake -DSIST_PLATFORM=arm64_linux -DSIST_DEBUG=off -DBUILD_TESTS=off -DCMAKE
 RUN make -j$(nproc)
 RUN strip sist2
 
-FROM --platform linux/arm64/v8 ubuntu:21.10
+FROM --platform="linux/arm64/v8" ubuntu:21.10
 
 RUN apt update && apt install -y curl libasan5 && rm -rf /var/lib/apt/lists/*
 

README.md

@@ -51,7 +51,7 @@ sist2 (Simple incremental search tool)
     1. Download the [latest sist2 release](https://github.com/simon987/sist2/releases) *
     1. *(or)* Download a [development snapshot](https://files.simon987.net/.gate/sist2/simon987_sist2/) *(Not
        recommended!)*
-    1. *(or)* `docker pull simon987/sist2:2.11.4-x64-linux`
+    1. *(or)* `docker pull simon987/sist2:2.11.6-x64-linux`
 
 1. See [Usage guide](docs/USAGE.md)
 
@@ -143,7 +143,7 @@ docker run --rm my-sist2-image cat /root/sist2 > sist2-x64-linux
 
     ```bash
     vcpkg install curl[core,openssl]
-    vcpkg install lmdb cjson glib brotli libarchive[core,bzip2,libxml2,lz4,lzma,lzo] pthread tesseract libxml2 libmupdf gtest mongoose libuuid libmagic libraw jasper lcms gumbo
+    vcpkg install lmdb cjson glib brotli libarchive[core,bzip2,libxml2,lz4,lzma,lzo] pthread tesseract libxml2 libmupdf gtest mongoose libmagic libraw jasper lcms gumbo
     ```
 
 1. Build

docs/USAGE.md

@@ -85,7 +85,7 @@ Exec-script options
 ### Scan options
 
 * `-t, --threads` 
-      Number of threads for file parsing. **Do not set a number higher than `$(nproc)` or `$(Get-WmiObject Win32_ComputerSystem).NumberOfLogicalProcessors` in Windows!**
+      Number of threads for file parsing. **Do not set a number higher than `$(nproc)` or `$(Get-CimInstance Win32_ComputerSystem).NumberOfLogicalProcessors` in Windows!**
 * `-q, --quality` 
     Thumbnail quality, on a scale of 1.0 to 31.0, 1.0 being the best.
 * `--size` 

src/cli.c

@@ -22,6 +22,7 @@
 const char *TESS_DATAPATHS[] = {
         "/usr/share/tessdata/",
         "/usr/share/tesseract-ocr/tessdata/",
+        "/usr/share/tesseract-ocr/4.00/tessdata/",
         "./",
         NULL
 };

src/io/walk.c

@@ -57,10 +57,11 @@ int handle_entry(const char *filepath, const struct stat *info, int typeflag, st
             pthread_mutex_lock(&ScanCtx.dbg_file_counts_mu);
             ScanCtx.dbg_excluded_files_count += 1;
             pthread_mutex_unlock(&ScanCtx.dbg_file_counts_mu);
-            return 0;
         } else if (typeflag == FTW_D) {
             return FTW_SKIP_SUBTREE;
         }
+
+        return FTW_CONTINUE;
     }
 
     if (typeflag == FTW_F && S_ISREG(info->st_mode)) {

src/log.c

@@ -55,10 +55,14 @@ void vsist_logf(const char *filepath, int level, char *format, va_list ap) {
         log_len += 1;
     }
 
-    int ret = write(STDERR_FILENO, log_str, log_len);
+    if (PrintingProgressBar) {
-    if (ret == -1) {
+        PrintingProgressBar = FALSE;
-        LOG_FATALF("serialize.c", "Could not write index descriptor: %s", strerror(errno))
+        memmove(log_str + 1, log_str, log_len);
+        log_str[0] = '\n';
+        log_len += 1;
     }
+
+    write(STDERR_FILENO, log_str, log_len);
 }
 
 void sist_logf(const char *filepath, int level, char *format, ...) {
@@ -104,8 +108,12 @@ void sist_log(const char *filepath, int level, char *str) {
         );
     }
 
-    int ret = write(STDERR_FILENO, log_str, log_len);
+    if (PrintingProgressBar) {
-    if (ret == -1) {
+        PrintingProgressBar = FALSE;
-        LOG_FATALF("serialize.c", "Could not write index descriptor: %s", strerror(errno));
+        memmove(log_str + 1, log_str, log_len);
+        log_str[0] = '\n';
+        log_len += 1;
     }
+
+    write(STDERR_FILENO, log_str, log_len);
 }

src/main.c

@@ -433,7 +433,7 @@ void sist2_index(index_args_t *args) {
         cleanup = elastic_cleanup;
     }
 
-    IndexCtx.pool = tpool_create(args->threads, cleanup, FALSE, TRUE);
+    IndexCtx.pool = tpool_create(args->threads, cleanup, FALSE, args->print == 0);
     tpool_start(IndexCtx.pool);
 
     struct dirent *de;
@@ -518,8 +518,8 @@ void sist2_web(web_args_t *args) {
 
 
 int main(int argc, const char *argv[]) {
-//    sigsegv_handler = signal(SIGSEGV, sig_handler);
+    sigsegv_handler = signal(SIGSEGV, sig_handler);
-//    sigabrt_handler = signal(SIGABRT, sig_handler);
+    sigabrt_handler = signal(SIGABRT, sig_handler);
 
     setlocale(LC_ALL, "");
 

src/sist.h

@@ -53,7 +53,7 @@
 #include <ctype.h>
 #include "git_hash.h"
 
-#define VERSION "2.11.4"
+#define VERSION "2.11.6"
 static const char *const Version = VERSION;
 
 #ifndef SIST_PLATFORM

src/tpool.c

@@ -191,7 +191,9 @@ void tpool_wait(tpool_t *pool) {
             }
         }
     }
+    if (pool->print_progress) {
         progress_bar_print(1.0, ScanCtx.stat_tn_size, ScanCtx.stat_index_size);
+    }
     pthread_mutex_unlock(&(pool->work_mutex));
 
     LOG_INFO("tpool.c", "Worker threads finished")

src/util.c

@@ -84,6 +84,8 @@ char *expandpath(const char *path) {
     return expanded;
 }
 
+int PrintingProgressBar = 0;
+
 void progress_bar_print(double percentage, size_t tn_size, size_t index_size) {
 
     static int last_val = -1;
@@ -114,13 +116,21 @@ void progress_bar_print(double percentage, size_t tn_size, size_t index_size) {
         index_unit = 'M';
     }
 
-    printf(
+    if (tn_size == 0 && index_size == 0) {
+        fprintf(stderr,
+                "\r%3d%%[%.*s>%*s]",
+                val, lpad, PBSTR, rpad, ""
+        );
+    } else {
+        fprintf(stderr,
                 "\r%3d%%[%.*s>%*s] TN:%3d%c IDX:%3d%c",
                 val, lpad, PBSTR, rpad, "",
                 (int) tn_size, tn_unit,
                 (int) index_size, index_unit
         );
-    fflush(stdout);
+    }
+
+    PrintingProgressBar = TRUE;
 }
 
 GHashTable *incremental_get_table() {

src/util.h

@@ -19,6 +19,8 @@ char *expandpath(const char *path);
 
 dyn_buffer_t url_escape(char *str);
 
+extern int PrintingProgressBar;
+
 void progress_bar_print(double percentage, size_t tn_size, size_t index_size);
 
 GHashTable *incremental_get_table();