fix(benchmark): restore supported formats, fix C# MIME mappings, bump version

Only remove pbm/pgm/ppm/pnm (truly unsupported), restore jp2/jpx/jpm/mj2,
typst/typ, and djot which are supported by kreuzberg core. Add missing
MIME type mappings in C# GuessMimeType for jp2, jpx, jpm, mj2, typst,
typ, djot, and markdown extensions.

Author: Goldziher

Cargo.lock

@@ -60,7 +60,7 @@ num_cpus = "1.17.0"
 once_cell = "1.21.3"
 parking_lot = "0.12.5"
 html-to-markdown-rs = { version = "2.24.5", default-features = false }
-reqwest = { version = "0.13.1", default-features = false }
+reqwest = { version = "0.13.2", default-features = false }
 image = { version = "0.25.9", default-features = false }
 clap = { version = "4.5", features = ["derive", "color", "suggestions"] }
 toml = "0.9.11"

Cargo.toml

@@ -126,7 +126,7 @@ simdutf8 = { version = "0.1", optional = true }
 hex = { workspace = true }
 lazy_static = "1.5.0"
 libc = { workspace = true }
-memchr = "2.7.6"
+memchr = "2.8.0"
 num_cpus = { workspace = true }
 once_cell = { workspace = true }
 parking_lot = { workspace = true }

crates/kreuzberg/Cargo.toml

@@ -220,7 +220,7 @@ static string GuessMimeType(string path)
         ".html" or ".htm" => "text/html",
         ".xml" => "application/xml",
         ".txt" => "text/plain",
-        ".md" or ".commonmark" => "text/markdown",
+        ".md" or ".markdown" or ".commonmark" => "text/markdown",
         ".csv" => "text/csv",
         ".tsv" => "text/tab-separated-values",
         ".rtf" => "application/rtf",
@@ -240,11 +240,17 @@ static string GuessMimeType(string path)
         ".gif" => "image/gif",
         ".bmp" => "image/bmp",
         ".webp" => "image/webp",
+        ".jp2" => "image/jp2",
+        ".jpx" => "image/jpx",
+        ".jpm" => "image/jpm",
+        ".mj2" => "video/mj2",
         ".svg" => "image/svg+xml",
         ".zip" => "application/zip",
         ".tar" => "application/x-tar",
         ".gz" or ".tgz" => "application/gzip",
         ".7z" => "application/x-7z-compressed",
+        ".typst" or ".typ" => "application/x-typst",
+        ".djot" => "text/djot",
         _ => "application/octet-stream",
     };
 }

packages/csharp/Benchmark/Program.cs

@@ -37,7 +37,7 @@ fn get_supported_formats(framework_name: &str) -> Vec<String> {
             "pdf",  // E-books
             "epub", // Vector/text
             "svg", "txt", // Images (for OCR) - gif and webp NOT supported by PyMuPDF
-            "png", "jpg", "jpeg", "bmp", "tiff", "tif", "pnm", "pgm", "pbm", "ppm",
+            "png", "jpg", "jpeg", "bmp", "tiff", "tif",
         ]
         .into_iter()
         .map(|s| s.to_string())
@@ -68,8 +68,7 @@ fn get_supported_formats(framework_name: &str) -> Vec<String> {
             "csv", "tsv", "json", "yaml", "yml", "toml", // Email
             "eml", "msg", // Scientific/technical (typst not supported - too new)
             "tex", "latex", "bib", "rst", "org", "ipynb", // Images (metadata + OCR)
-            "png", "jpg", "jpeg", "gif", "bmp", "tiff", "tif", "webp", "jp2", "pnm", "pgm", "pbm", "ppm",
-            // Archives
+            "png", "jpg", "jpeg", "gif", "bmp", "tiff", "tif", "webp", "jp2", // Archives
             "zip", "tar", "gz", "7z",
         ]
         .into_iter()

tools/benchmark-harness/src/adapters/external.rs

@@ -72,14 +72,21 @@ fn get_kreuzberg_supported_formats() -> Vec<String> {
         "tiff",
         "tif",
         "webp",
+        "jp2",
+        "jpx",
+        "jpm",
+        "mj2",
         // Academic/Publishing
         "epub",
         "bib",
         "ipynb",
         "tex",
         "latex",
+        "typst",
+        "typ",
         // Other
         "svg",
+        "djot",
     ]
     .into_iter()
     .map(|s| s.to_string())

tools/benchmark-harness/src/adapters/kreuzberg.rs

@@ -110,7 +110,7 @@ impl FrameworkAdapter for NativeAdapter {
             "zip" | "tar" | "gz" | "tgz" | "7z" |
             // Images (OCR supported)
             "bmp" | "gif" | "jpg" | "jpeg" | "png" | "tiff" | "tif" | "webp" |
-            "jp2" | "jpx" | "jpm" | "mj2" | "pnm" | "pbm" | "pgm" | "ppm" |
+            "jp2" | "jpx" | "jpm" | "mj2" |
             // Academic/Publishing
             "epub" | "bib" | "ipynb" | "tex" | "latex" | "typst" | "typ" |
             // Other