Merge pull request #30 from jmattaa/feat/file-extensions

feat: add file extensions to filetype detection

include/filetypes/checktypes.h

@@ -3,18 +3,27 @@
 
 #include <stddef.h>
 
-struct laser_magicnumber
+// https://en.wikipedia.org/wiki/List_of_file_signatures
+// dis be good sometimes
+
+struct laser_filetype
 {
     unsigned char *magic;
     size_t magic_size;
+
+    const char **extensions;
+    size_t extensions_count;
 };
 
 int laser_checktype(const char *filename,
-                    const struct laser_magicnumber formats[]);
-int laser_checktype_ex(int fd, const char *filename,
-                       const struct laser_magicnumber formats[]);
-extern const struct laser_magicnumber laser_archiveformats[];
-extern const struct laser_magicnumber laser_mediaformats[];
-extern const struct laser_magicnumber laser_documentformats[];
+                    const struct laser_filetype formats[]);
+
+int laser_checktype_extension(const char *filename,
+                              const struct laser_filetype formats[]);
+int laser_checktype_magic(int fd, const struct laser_filetype formats[]);
+
+extern const struct laser_filetype laser_archiveformats[];
+extern const struct laser_filetype laser_mediaformats[];
+extern const struct laser_filetype laser_documentformats[];
 
 #endif

src/filetypes/archives.c

@@ -1,14 +1,26 @@
 #include "filetypes/checktypes.h"
 
-const struct laser_magicnumber laser_archiveformats[] = {
-    {(unsigned char[]){0x50, 0x4B, 0x03, 0x04}, 4}, // ZIP and stuff based on it
-    {(unsigned char[]){0x1F, 0x8B}, 2},             // GZIP
-    {(unsigned char[]){'u', 's', 't', 'a', 'r'}, 5},      // TAR
-    {(unsigned char[]){0x42, 0x5A, 0x68}, 3},             // BZIP2
-    {(unsigned char[]){0xFD, 0x37, 0x7A, 0x58, 0x5A}, 5}, // XZ
-    {(unsigned char[]){0x52, 0x61, 0x72, 0x21, 0x1A, 0x07, 0x00},
-     7}, // RAR v1.5 to 4.0
-    {(unsigned char[]){0x52, 0x61, 0x72, 0x21, 0x1A, 0x07, 0x01, 0x00},
-     8}, // RAR v5+
+const struct laser_filetype laser_archiveformats[] = {
+
+    {
+        (unsigned char[]){0x50, 0x4B, 0x03, 0x04},
+        4,
+        (const char *[]){"zip", "aar",  "apk",  "docx", "epub", "ipa", "jar",
+                         "kmz", "maff", "msix", "odp",  "ods",  "odt", "pk3",
+                         "pk4", "pptx", "usdz", "vsdx", "xlsx", "xpi"},
+        20,
+    }, // ZIP and stuff based on it
+    {(unsigned char[]){0x1F, 0x8B}, 2, (const char *[]){"gz"}, 1}, // GZIP
+    {(unsigned char[]){'u', 's', 't', 'a', 'r'}, 5, (const char *[]){"tar"},
+     1}, // TAR
+    {(unsigned char[]){0x42, 0x5A, 0x68}, 3, (const char *[]){"bz2"},
+     1}, // BZIP2
+    {(unsigned char[]){0xFD, 0x37, 0x7A, 0x58, 0x5A}, 5, (const char *[]){"xz"},
+     1}, // XZ
+    {(unsigned char[]){0x52, 0x61, 0x72, 0x21, 0x1A, 0x07, 0x00}, 7,
+     (const char *[]){"rar"}, 1}, // RAR v1.5 to 4.0
+    {(unsigned char[]){0x52, 0x61, 0x72, 0x21, 0x1A, 0x07, 0x01, 0x00}, 8,
+     (const char *[]){"rar"}, 1}, // RAR v5+
     // ADD MORE ARCHIVES
-    {NULL, 0}};
+
+    {NULL, 0, NULL, 0}};

src/filetypes/checktype.c

@@ -12,14 +12,68 @@ enum check_type_return
     CannotRead,
     Unknown
 };
-static int laser_checktype_fd(int fd, const struct laser_magicnumber formats[])
+
+int laser_checktype(const char *filename, const struct laser_filetype formats[])
+{
+    if (laser_checktype_extension(filename, formats) == Success)
+        return 1;
+
+    int fd = open(filename, O_RDONLY);
+    if (fd == -1)
+    {
+        perror("open");
+        return 0;
+    }
+
+    int rv = laser_checktype_magic(fd, formats);
+    switch (rv)
+    {
+        case NotOpened:
+            perror("open");
+            break;
+        case CannotRead:
+            fprintf(stderr, "lsr: cannot read %s", filename);
+            break;
+        default:
+            break;
+    }
+    close(fd);
+    return rv == Success;
+}
+
+int laser_checktype_extension(const char *filename,
+                              const struct laser_filetype *formats)
+{
+    int rv = Unknown;
+    for (int i = 0; formats[i].extensions != NULL; i++)
+    {
+        for (size_t j = 0; j < formats[i].extensions_count; j++)
+        {
+            char *ext = strrchr(filename, '.') + 1; // +1 skip .
+            if (strcmp(ext, formats[i].extensions[j]) == 0)
+                rv = Success;
+        }
+    }
+    return rv;
+}
+
+int laser_checktype_magic(int fd, const struct laser_filetype formats[])
 {
+
+    if (fd == -1)
+    {
+        perror("open");
+        return 0;
+    }
+
     off_t curr_pos = lseek(fd, 0, SEEK_CUR);
     if (fd == -1)
     {
         return NotOpened;
     }
     lseek(fd, 0, SEEK_SET); // seek to start
+
+    // ----------------------------------------------------------------------
     // should be updated if there is a magic number larger than 8 bits
     unsigned char buffer[8];
 
@@ -46,52 +100,3 @@ static int laser_checktype_fd(int fd, const struct laser_magicnumber formats[])
     lseek(fd, curr_pos, SEEK_SET);
     return Unknown;
 }
-
-int laser_checktype(const char *filename,
-                    const struct laser_magicnumber formats[])
-{
-    int fd = open(filename, O_RDONLY);
-    if (fd == -1)
-    {
-        perror("open");
-        return 0;
-    }
-    int rv = laser_checktype_fd(fd, formats);
-    switch (rv)
-    {
-        case NotOpened:
-            perror("open");
-            break;
-        case CannotRead:
-            fprintf(stderr, "laser: cannot read %s", filename);
-            break;
-        default:
-            break;
-    }
-    close(fd);
-    return rv == Success;
-}
-
-int laser_checktype_ex(int fd, const char *filename,
-                       const struct laser_magicnumber formats[])
-{
-
-    if (fd == -1)
-    {
-        perror("open");
-        return 0;
-    }
-    int rv = laser_checktype_fd(fd, formats);
-    switch (rv)
-    {
-        case NotOpened:
-            perror("open");
-            break;
-        case CannotRead:
-            fprintf(stderr, "laser: cannot read %s", filename);
-            break;
-        default:
-            break;
-    }
-    return rv == Success;
-}

src/filetypes/documents.c

@@ -1,10 +1,13 @@
 #include "filetypes/checktypes.h"
 
-const struct laser_magicnumber laser_documentformats[] = {
-    {(unsigned char[]){'%', 'P', 'D', 'F'}, 4}, // PDF
-    {(unsigned char[]){0xD0, 0xCF, 0x11, 0xE0, 0xA1, 0xB1, 0x1A, 0xE1},
-     8}, // compound bin file from microsoft (doc, xls, ppt, msi, msg)
-    {(unsigned char[]){0x41, 0x54, 0x26, 0x54, 0x46, 0x4F, 0x52, 0x4D},
-     8}, // DJVU
+const struct laser_filetype laser_documentformats[] = {
+    {(unsigned char[]){'%', 'P', 'D', 'F'}, 4, (const char *[]){"pdf"},
+     1}, // PDF
+    {(unsigned char[]){0xD0, 0xCF, 0x11, 0xE0, 0xA1, 0xB1, 0x1A, 0xE1}, 8,
+     (const char *[]){"doc", "xls", "ppt", "msi", "msg"},
+     5}, // compound bin file from microsoft
+    {(unsigned char[]){0x41, 0x54, 0x26, 0x54, 0x46, 0x4F, 0x52, 0x4D}, 8,
+     (const char *[]){"djvu", "djv"}, 2}, // DJVU
     // ADD DOCUMENT TYPES
-    {NULL, 0}};
+
+    {NULL, 0, NULL, 0}};

src/filetypes/medias.c

@@ -1,11 +1,15 @@
 #include "filetypes/checktypes.h"
 
-const struct laser_magicnumber laser_mediaformats[] = {
-    {(unsigned char[]){0xFF, 0xD8, 0xFF, 0xE0}, 4}, // JPEG
-    {(unsigned char[]){0x89, 0x50, 0x4E, 0x47}, 4}, // PNG
-    {(unsigned char[]){0x47, 0x49, 0x46, 0x38}, 4}, // GIF
-    {(unsigned char[]){0x49, 0x44, 0x33}, 3},       // MP3
-    {(unsigned char[]){0x66, 0x74, 0x79, 0x70, 0x4D, 0x53, 0x4E, 0x56},
-     8}, // MP4
+const struct laser_filetype laser_mediaformats[] = {
+    {(unsigned char[]){0xFF, 0xD8, 0xFF, 0xE0}, 4,
+     (const char *[]){"jpeg", "jpg"}, 2}, // JPEG
+    {(unsigned char[]){0x89, 0x50, 0x4E, 0x47}, 4, (const char *[]){"png"},
+     1}, // PNG
+    {(unsigned char[]){0x47, 0x49, 0x46, 0x38}, 4, (const char *[]){"gif"},
+     1},                                                                  // GIF
+    {(unsigned char[]){0x49, 0x44, 0x33}, 3, (const char *[]){"mp3"}, 1}, // MP3
+    {(unsigned char[]){0x66, 0x74, 0x79, 0x70, 0x4D, 0x53, 0x4E, 0x56}, 8,
+     (const char *[]){"mp4"}, 1}, // MP4
     // ADD MORE MEDIA TYPES
-    {NULL, 0}};
+
+    {NULL, 0, NULL, 0}};

src/laser.c

@@ -98,15 +98,13 @@ void laser_print_entry(struct laser_dirent *entry, const char *color,
 
 static laser_color_type laser_color_for_format(const char *filename)
 {
-    int fd = open(filename, O_RDONLY);
     laser_color_type type = LASER_COLOR_FILE;
-    if (laser_checktype_ex(fd, filename, laser_archiveformats))
+    if (laser_checktype(filename, laser_archiveformats))
         type = LASER_COLOR_ARCHIVE;
-    else if (laser_checktype_ex(fd, filename, laser_mediaformats))
+    else if (laser_checktype(filename, laser_mediaformats))
         type = LASER_COLOR_MEDIA;
-    else if (laser_checktype_ex(fd, filename, laser_documentformats))
+    else if (laser_checktype(filename, laser_documentformats))
         type = LASER_COLOR_DOCUMENT;
-    close(fd);
     return type;
 }