arc component

.gitignore

@@ -5,4 +5,5 @@ libscan.a
 libscan.so
 *.cbp
 CMakeFiles
-CMakeCache.txt
+CMakeCache.txt
+scan_test

.gitmodules

@@ -4,3 +4,9 @@
 [submodule "third-party/utf8.h"]
 	path = third-party/utf8.h
 	url = https://github.com/sheredom/utf8.h
+[submodule "third-party/libarchive"]
+	path = third-party/libarchive
+	url = https://github.com/libarchive/libarchive
+[submodule "third-party/zlib"]
+	path = third-party/zlib
+	url = https://github.com/madler/zlib

CMakeLists.txt

@@ -1,33 +1,45 @@
 cmake_minimum_required(VERSION 3.15)
-project(scan C)
 
+project(scan C)
 set(CMAKE_C_STANDARD 11)
 
+find_package(LibArchive REQUIRED)
 
 add_library(
         scan
-        src/text.c src/text.h
-        src/util.c src/util.h
-        src/types.h
-        src/macros.h
+        libscan/util.c libscan/util.h
+        libscan/scan.c libscan/scan.h
+        libscan/macros.h
+
+        libscan/text/text.c libscan/text/text.h
+        libscan/arc/arc.c libscan/arc/arc.h
 
         third-party/utf8.h
 )
 
-
-# Third party
-add_subdirectory(third-party/uuid/)
-target_include_directories(
-        scan PRIVATE
-        third-party/uuid/src/
-)
-
 target_compile_options(
-        scan PRIVATE
+        scan
+        PRIVATE
         -Werror
+        -g
 )
 
 target_link_libraries(
         scan
-        uuid
-)
+        -static
+        ${LibArchive_LIBRARIES}
+)
+
+
+# test executable
+add_executable(
+        scan_test
+        test/main.c
+)
+
+
+target_link_libraries(
+        scan_test
+        scan
+)
+

README.md

@@ -0,0 +1,4 @@
+
+```bash
+vcpkg install libarchive
+```

libscan/arc/arc.c

@@ -0,0 +1,167 @@
+#include "arc.h"
+
+#include "../scan.h"
+#include "../util.h"
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <fcntl.h>
+
+
+
+int should_parse_filtered_file(const char *filepath, int ext) {
+    char tmp[PATH_MAX * 2];
+
+    if (ext == 0) {
+        return FALSE;
+    }
+
+    memcpy(tmp, filepath, ext - 1);
+    *(tmp + ext - 1) = '\0';
+
+    char *idx = strrchr(tmp, '.');
+
+    if (idx == NULL) {
+        return FALSE;
+    }
+
+    if (strcmp(idx, ".tar") == 0) {
+        return TRUE;
+    }
+
+    return FALSE;
+}
+
+int arc_read(struct vfile *f, void *buf, size_t size) {
+    return archive_read_data(f->arc, buf, size);
+}
+
+typedef struct arc_data {
+    vfile_t *f;
+    char buf[ARC_BUF_SIZE];
+} arc_data_f;
+
+int vfile_open_callback(struct archive *a, void *user_data) {
+    arc_data_f *data = user_data;
+
+    if (data->f->is_fs_file && data->f->fd == -1) {
+        data->f->fd = open(data->f->filepath, O_RDONLY);
+    }
+
+    return ARCHIVE_OK;
+}
+
+long vfile_read_callback(struct archive *a, void *user_data, const void **buf) {
+    arc_data_f *data = user_data;
+
+    *buf = data->buf;
+    return data->f->read(data->f, data->buf, ARC_BUF_SIZE);
+}
+
+int vfile_close_callback(struct archive *a, void *user_data) {
+    arc_data_f *data = user_data;
+
+    if (data->f->close != NULL) {
+        data->f->close(data->f);
+    }
+
+    return ARCHIVE_OK;
+}
+
+scan_code_t parse_archive(scan_arc_ctx_t *ctx, vfile_t *f, document_t *doc) {
+
+    struct archive *a;
+    struct archive_entry *entry;
+
+
+    arc_data_f data;
+    data.f = f;
+
+    int ret = 0;
+    if (data.f->is_fs_file) {
+
+        a = archive_read_new();
+        archive_read_support_filter_all(a);
+        archive_read_support_format_all(a);
+
+        ret = archive_read_open_filename(a, doc->filepath, ARC_BUF_SIZE);
+    } else if (ctx->mode == ARC_MODE_RECURSE) {
+
+        a = archive_read_new();
+        archive_read_support_filter_all(a);
+        archive_read_support_format_all(a);
+
+        ret = archive_read_open(
+                a, &data,
+                vfile_open_callback,
+                vfile_read_callback,
+                vfile_close_callback
+        );
+    } else {
+        return SCAN_OK;
+    }
+
+    if (ret != ARCHIVE_OK) {
+        //TODO: log
+//        LOG_ERRORF(doc->filepath, "(arc.c) [%d] %s", ret, archive_error_string(a))
+        archive_read_free(a);
+        return SCAN_ERR_READ;
+    }
+
+    if (ctx->mode == ARC_MODE_LIST) {
+
+        dyn_buffer_t buf = dyn_buffer_create();
+
+        while (archive_read_next_header(a, &entry) == ARCHIVE_OK) {
+            if (S_ISREG(archive_entry_stat(entry)->st_mode)) {
+
+                char *path = (char *) archive_entry_pathname(entry);
+
+                dyn_buffer_append_string(&buf, path);
+                dyn_buffer_write_char(&buf, '\n');
+            }
+        }
+        dyn_buffer_write_char(&buf, '\0');
+
+        meta_line_t *meta_list = malloc(sizeof(meta_line_t) + buf.cur);
+        meta_list->key = MetaContent;
+        strcpy(meta_list->str_val, buf.buf);
+        APPEND_META(doc, meta_list);
+        dyn_buffer_destroy(&buf);
+
+    } else {
+
+        parse_job_t *sub_job = malloc(sizeof(parse_job_t) + PATH_MAX * 2);
+
+        sub_job->vfile.close = NULL;
+        sub_job->vfile.read = arc_read;
+        sub_job->vfile.arc = a;
+        sub_job->vfile.filepath = sub_job->filepath;
+        sub_job->vfile.is_fs_file = FALSE;
+        memcpy(sub_job->parent, doc->uuid, sizeof(uuid_t));
+
+        while (archive_read_next_header(a, &entry) == ARCHIVE_OK) {
+            sub_job->info = *archive_entry_stat(entry);
+            if (S_ISREG(sub_job->info.st_mode)) {
+                sprintf(sub_job->filepath, "%s#/%s", f->filepath, archive_entry_pathname(entry));
+                sub_job->base = (int) (strrchr(sub_job->filepath, '/') - sub_job->filepath) + 1;
+
+                char *p = strrchr(sub_job->filepath, '.');
+                if (p != NULL) {
+                    sub_job->ext = (int) (p - sub_job->filepath + 1);
+                } else {
+                    sub_job->ext = (int) strlen(sub_job->filepath);
+                }
+
+                //TODO:
+//                parse(sub_job);
+            }
+        }
+
+        free(sub_job);
+    }
+
+    archive_read_free(a);
+    return SCAN_OK;
+}

libscan/arc/arc.h

@@ -0,0 +1,26 @@
+#ifndef SCAN_ARC_H
+#define SCAN_ARC_H
+
+#include <archive.h>
+#include <archive_entry.h>
+#include "../scan.h"
+
+#define ARC_MODE_SKIP 0
+#define ARC_MODE_LIST 1
+#define ARC_MODE_SHALLOW 2
+#define ARC_MODE_RECURSE 3
+typedef int archive_mode_t;
+
+typedef struct {
+    archive_mode_t mode;
+} scan_arc_ctx_t;
+
+#define ARC_BUF_SIZE 8192
+
+int should_parse_filtered_file(const char *filepath, int ext);
+
+scan_code_t parse_archive(scan_arc_ctx_t *ctx, vfile_t *f, document_t *doc);
+
+int arc_read(struct vfile * f, void *buf, size_t size);
+
+#endif

libscan/macros.h

@@ -11,3 +11,7 @@
 
 #undef	MIN
 #define MIN(a, b)  (((a) < (b)) ? (a) : (b))
+
+#ifndef PATH_MAX
+#define PATH_MAX 4096
+#endif

libscan/scan.c

@@ -0,0 +1,24 @@
+#include "scan.h"
+
+#include <fcntl.h>
+#include <unistd.h>
+
+int fs_read(struct vfile *f, void *buf, size_t size) {
+    if (f->fd == -1) {
+        f->fd = open(f->filepath, O_RDONLY);
+        if (f->fd == -1) {
+            //TODO: log
+//            LOG_ERRORF(f->filepath, "open(): [%d] %s", errno, strerror(errno))
+            return -1;
+        }
+    }
+
+    return read(f->fd, buf, size);
+}
+
+
+void fs_close(struct vfile *f) {
+    if (f->fd != -1) {
+        close(f->fd);
+    }
+}

libscan/scan.h

@@ -1,18 +1,11 @@
-#ifndef SCAN_TYPES_H
-#define SCAN_TYPES_H
+#ifndef SCAN_SCAN_H
+#define SCAN_SCAN_H
 
 #include <stdio.h>
 #include <sys/stat.h>
-#include "uuid.h"
+#include "../third-party/uuid/src/uuid.h"
 
 
-// TODO
-#define ARC_MODE_SKIP 0
-#define ARC_MODE_LIST 1
-#define ARC_MODE_SHALLOW 2
-#define ARC_MODE_RECURSE 3
-typedef int archive_mode_t;
-
 #define META_INT_MASK 0x80
 #define META_STR_MASK 0x40
 #define META_LONG_MASK 0x20
@@ -121,4 +114,11 @@ typedef struct parse_job_t {
         doc->meta_tail = meta;\
     }
 
+
 #endif
+
+void fs_close(struct vfile *f);
+
+#define CLOSE_FILE(f) if (f.close != NULL) {f.close(&f);};
+
+int fs_read(struct vfile *f, void *buf, size_t size);

libscan/text/text.c

@@ -21,6 +21,8 @@ scan_code_t parse_text(scan_text_ctx_t *ctx, struct vfile *f, document_t *doc) {
 
     APPEND_META(doc, meta)
 
+    printf("%s", meta->str_val);
+
     free(buf);
     text_buffer_destroy(&tex);
 

libscan/text/text.h

@@ -1,8 +1,8 @@
 #ifndef SCAN_TEXT_H
 #define SCAN_TEXT_H
 
-#include "types.h"
-#include "util.h"
+#include "../scan.h"
+#include "../util.h"
 
 typedef struct {
     long content_size;

test/main.c

@@ -0,0 +1,22 @@
+#include "../libscan/text/text.h"
+#include <fcntl.h>
+#include "../libscan/arc/arc.h"
+
+int main() {
+
+    scan_text_ctx_t ctx;
+
+    ctx.content_size = 100;
+    vfile_t file;
+    file.is_fs_file = TRUE;
+    file.filepath = "/home/simon/Downloads/libscan/CMakeLists.txt";
+    file.fd = open("/home/simon/Downloads/libscan/CMakeLists.txt", O_RDONLY);
+    file.read = fs_read;
+
+    document_t doc;
+    doc.meta_head = NULL;
+    doc.meta_tail = NULL;
+
+    doc.size = 200;
+    parse_text(&ctx, &file, &doc);
+}