Add a Mimetype_Tester, some more mimetype stuff.

kohler · Sep 7, 2023 · 28aeff6 · 28aeff6
1 parent c5807d3
commit 28aeff6
Show file tree

Hide file tree

Showing 7 changed files with 136 additions and 44 deletions.
diff --git a/batch/updatedocmetadata.php b/batch/updatedocmetadata.php
@@ -44,7 +44,7 @@ private function run_images() {
     private function run_images_subset($docs) {
         DocumentInfo::prefetch_content($docs, DocumentInfo::FLAG_NO_DOCSTORE);
         foreach ($docs as $doc) {
-            $info = Mimetype::content_info($doc->content(), $doc->mimetype);
+            $info = Mimetype::content_info(null, $doc->mimetype, $doc);
             $upd = [];
             $m = $doc->metadata() ?? (object) [];
             if (isset($info["width"]) && !isset($m->width)) {

diff --git a/lib/isovideomimetype.php b/lib/isovideomimetype.php
@@ -31,6 +31,7 @@ class ISOVideoMimetype implements JsonSerializable {
     const F_FTYP = 1;
     const F_MOOV = 2;
     const F_MVEX = 4;
+    const F_ANALYZED = 8;
 
     const TF_AUDIO = 1;
     const TF_VIDEO = 2;
@@ -41,20 +42,21 @@ private function __construct() {
     }
 
     /** @param string $s
-     * @return VideoMimetype */
+     * @return ISOVideoMimetype */
     static function make_string($s) {
-        $vm = new VideoMimetype;
+        $vm = new ISOVideoMimetype;
         $vm->data = new ISOVideoFragment($s, 0, strlen($s));
         $vm->bound = strlen($s);
         return $vm;
     }
 
     /** @param string $filename
-     * @return VideoMimetype */
-    static function make_file($filename) {
-        $vm = new VideoMimetype;
+     * @param ?string $prefix
+     * @return ISOVideoMimetype */
+    static function make_file($filename, $prefix = null) {
+        $vm = new ISOVideoMimetype;
         $vm->filename = $filename;
-        $s = file_get_contents($filename, false, null, 0, 32768);
+        $s = $prefix ?? file_get_contents($filename, false, null, 0, 32768);
         $vm->data = new ISOVideoFragment($s, 0, strlen($s));
         $vm->bound = @filesize($filename);
         return $vm;
@@ -272,7 +274,12 @@ function walk_hdlr($data, $pos, $bound, $track) {
     }
 
 
-    function walk_boxes() {
+    function analyze() {
+        if (($this->flags & self::F_ANALYZED) !== 0) {
+            return;
+        }
+        $this->flags |= self::F_ANALYZED;
+
         $data = $this->data;
         $pos = 0;
         while (($this->flags & 3) !== 3
@@ -324,10 +331,10 @@ function walk_boxes() {
                     $this->tflags |= self::TF_ASPECT;
                 } else {
                     if ($this->width === null && $tr->width !== null) {
-                        $this->width = $tr->width / 65536.0;
+                        $this->width = (int) round($tr->width / 65536.0);
                     }
                     if ($this->height === null && $tr->height !== null) {
-                        $this->height = $tr->height / 65536.0;
+                        $this->height = (int) round($tr->height / 65536.0);
                     }
                 }
             } else if ($tr->handler === 0x736f756e /* `soun` */) {
@@ -343,6 +350,31 @@ function walk_boxes() {
         }
     }
 
+    /** @param ?string $type
+     * @return array */
+    function content_info($type = null) {
+        $this->analyze();
+        if ($this->tflags === 0) {
+            return $type ? ["type" => $type] : [];
+        }
+        if ($type === null) {
+            if (($this->tflags & self::TF_VIDEO) !== 0) {
+                $type = "video/mp4";
+            } else {
+                $type = "audio/mp4";
+            }
+        }
+        $info = ["type" => $type];
+        if ($this->duration !== null) {
+            $info["duration"] = $this->duration;
+        }
+        if ($this->width !== null && $this->height !== null) {
+            $info["width"] = $this->width;
+            $info["height"] = $this->height;
+        }
+        return $info;
+    }
+
     #[\ReturnTypeWillChange]
     function jsonSerialize() {
         $j = ["flags" => $this->flags];

diff --git a/lib/mimetype.php b/lib/mimetype.php
@@ -309,8 +309,19 @@ static function content_type($content, $type = null) {
 
     /** @param ?string $content
      * @param ?string $type
-     * @return array{type:string,width?:int,height?:int} */
-    static function content_info($content, $type = null) {
+     * @param ?DocumentInfo $doc
+     * @return ?array{type:string,width?:int,height?:int} */
+    static function content_info($content, $type = null, $doc = null) {
+        if ($content === null && $doc) {
+            if ($doc->has_memory_content()) {
+                $content = $doc->content();
+            } else {
+                $content = $doc->content_prefix(4096);
+            }
+            if ($content === false) {
+                return null;
+            }
+        }
         $content = $content ?? "";
         $type = self::content_type($content, $type);
         if ($type === self::JPG_TYPE) {
@@ -319,6 +330,15 @@ static function content_info($content, $type = null) {
             return self::png_content_info($content);
         } else if ($type === self::GIF_TYPE) {
             return self::gif_content_info($content);
+        } else if ($type === "video/mp4") {
+            if ($doc
+                && strlen($content) !== $doc->size()
+                && ($file = $doc->content_file())) {
+                $ivm = ISOVideoMimetype::make_file($file, $content);
+            } else {
+                $ivm = ISOVideoMimetype::make_string($content);
+            }
+            return $ivm->content_info();
         } else {
             return ["type" => $type];
         }

diff --git a/src/documentinfo.php b/src/documentinfo.php
@@ -311,17 +311,20 @@ static function sanitize_filename($fn) {
     }
 
     function analyze_content() {
-        $pfx = $this->content_prefix(4096);
-        if ($pfx === false) {
-            return;
-        }
-        $info = Mimetype::content_info($pfx, $this->mimetype);
-        $this->mimetype = $info["type"];
-        if (isset($info["width"])) {
-            $this->set_prop("width", $info["width"]);
-        }
-        if (isset($info["height"])) {
-            $this->set_prop("height", $info["height"]);
+        if (($info = Mimetype::content_info(null, $this->mimetype, $this))) {
+            $this->mimetype = $info["type"];
+            if (isset($info["width"])) {
+                $this->set_prop("width", $info["width"]);
+            }
+            if (isset($info["height"])) {
+                $this->set_prop("height", $info["height"]);
+            }
+            if (str_starts_with($this->mimetype, "video/")
+                || str_starts_with($this->mimetype, "audio/")) {
+                if (isset($info["duration"])) {
+                    $this->set_prop("npages", (int) ($info["duration"] * 10 + 0.5));
+                }
+            }
         }
     }
 
@@ -984,6 +987,11 @@ private function _temp_content_filename() {
         return Filer::$tempdir . "/" . $base . Mimetype::extension($this->mimetype);
     }
 
+    /** @return bool */
+    function has_memory_content() {
+        return $this->content !== null;
+    }
+
     /** @param int $prefix_len
      * @return string|false */
     function content_prefix($prefix_len = 4096) {

diff --git a/test/t_documentbasics.php b/test/t_documentbasics.php
@@ -97,24 +97,4 @@ function test_docstore_path() {
         xassert_eqq(Filer::docstore_path($doc), "/foo/bar/sha2-66a/sha2-66a04/sha2-66a045b452102c59d840ec097d59d9467e13a3f34f6494e539ffd32c1bb35f18");
         xassert_eqq($doc->s3_key(), "doc/66a/sha2-66a045b452102c59d840ec097d59d9467e13a3f34f6494e539ffd32c1bb35f18.txt");
     }
-
-    function test_mimetype() {
-        xassert_eqq(Mimetype::content_type("%PDF-3.0\nwhatever\n"), Mimetype::PDF_TYPE);
-        // test that we can parse lib/mime.types for file extensions
-        xassert_eqq(Mimetype::extension("application/pdf"), ".pdf");
-        xassert_eqq(Mimetype::extension("image/gif"), ".gif");
-        xassert_eqq(Mimetype::content_type(null, "application/force"), "application/octet-stream");
-        xassert_eqq(Mimetype::content_type(null, "application/x-zip-compressed"), "application/zip");
-        xassert_eqq(Mimetype::content_type(null, "application/gz"), "application/gzip");
-        xassert_eqq(Mimetype::extension("application/g-zip"), ".gz");
-        xassert_eqq(Mimetype::type("application/download"), "application/octet-stream");
-        xassert_eqq(Mimetype::extension("application/smil"), ".smil");
-        xassert_eqq(Mimetype::type(".smil"), "application/smil");
-        xassert_eqq(Mimetype::type(".sml"), "application/smil");
-        // `fileinfo` test
-        xassert_eqq(Mimetype::content_type("<html><head></head><body></body></html>"), "text/html");
-        // test that non-PDFs are not mistaken for PDFs
-        xassert_eqq(Mimetype::content_type("%PDF-3.0\nwhatever\n", Mimetype::PDF_TYPE), Mimetype::PDF_TYPE);
-        xassert_neqq(Mimetype::content_type("PDF-3.0\nwhatever\n", Mimetype::PDF_TYPE), Mimetype::PDF_TYPE);
-    }
 }
diff --git a/test/t_mimetype.php b/test/t_mimetype.php
@@ -0,0 +1,51 @@
+<?php
+// t_mimetype.php -- HotCRP tests
+// Copyright (c) 2006-2023 Eddie Kohler; see LICENSE.
+
+class Mimetype_Tester {
+    function test_mimetype() {
+        xassert_eqq(Mimetype::content_type("%PDF-3.0\nwhatever\n"), Mimetype::PDF_TYPE);
+        // test that we can parse lib/mime.types for file extensions
+        xassert_eqq(Mimetype::extension("application/pdf"), ".pdf");
+        xassert_eqq(Mimetype::extension("image/gif"), ".gif");
+        xassert_eqq(Mimetype::content_type(null, "application/force"), "application/octet-stream");
+        xassert_eqq(Mimetype::content_type(null, "application/x-zip-compressed"), "application/zip");
+        xassert_eqq(Mimetype::content_type(null, "application/gz"), "application/gzip");
+        xassert_eqq(Mimetype::extension("application/g-zip"), ".gz");
+        xassert_eqq(Mimetype::type("application/download"), "application/octet-stream");
+        xassert_eqq(Mimetype::extension("application/smil"), ".smil");
+        xassert_eqq(Mimetype::type(".smil"), "application/smil");
+        xassert_eqq(Mimetype::type(".sml"), "application/smil");
+        // `fileinfo` test
+        xassert_eqq(Mimetype::content_type("<html><head></head><body></body></html>"), "text/html");
+        // test that non-PDFs are not mistaken for PDFs
+        xassert_eqq(Mimetype::content_type("%PDF-3.0\nwhatever\n", Mimetype::PDF_TYPE), Mimetype::PDF_TYPE);
+        xassert_neqq(Mimetype::content_type("PDF-3.0\nwhatever\n", Mimetype::PDF_TYPE), Mimetype::PDF_TYPE);
+    }
+
+    function test_gif() {
+        $spacer = base64_decode("R0lGODlhAQABAIAAAAAAAAAAACH5BAEAAAAALAAAAAABAAEAAAICRAEAOw==");
+        xassert_eqq(Mimetype::content_type($spacer), Mimetype::GIF_TYPE);
+        $ci = Mimetype::content_info($spacer);
+        xassert_eqq($ci["type"], Mimetype::GIF_TYPE);
+        xassert_eqq($ci["width"] ?? null, 1);
+        xassert_eqq($ci["height"] ?? null, 1);
+
+        $icon = base64_decode("R0lGODlhVQBVAPQAAPcIWgBrMf/WtZSMSmMYKfelUggACPdSlK0AQu+MxvecUlJ7OfcQY/+1hPcYa0IAGOecWu+11oQAMfd7lLWUSiFzOecIUvc5c2uEQv+ErTl7OcYASvcha//OjPelYwAAACH5BAEAABEALAAAAABVAFUAQAU=");
+        xassert_eqq(Mimetype::content_type($icon), Mimetype::GIF_TYPE);
+        $ci = Mimetype::content_info($icon);
+        xassert_eqq($ci["type"], Mimetype::GIF_TYPE);
+        xassert_eqq($ci["width"] ?? null, 85);
+        xassert_eqq($ci["height"] ?? null, 85);
+    }
+
+    function xxx_test_mp4() {
+        $mt = ISOVideoMimetype::make_file("/Users/kohler/Downloads/sigcomm23-paper130-10_minute_presentation_video.mp4");
+        $mt->analyze();
+        error_log(json_encode($mt->content_info()));
+
+        $mt = ISOVideoMimetype::make_file("/Users/kohler/Downloads/sigcomm23-paper1037-10_minute_presentation_video/MoMA v1.0.3.mp4");
+        $mt->analyze();
+        error_log(json_encode($mt->content_info()));
+    }
+}
diff --git a/test/test02.php b/test/test02.php
@@ -18,5 +18,6 @@
     "Search_Tester",
     "Settings_Tester",
     "UpdateSchema_Tester",
-    "Batch_Tester"
+    "Batch_Tester",
+    "Mimetype_Tester"
 );