Advanced Topics
This section covers advanced usage and internals of Apache Tika.
Most pages here are written from a Java-API perspective. Where a topic
has a JSON-config or CLI equivalent, look first under
Configuration (per-parser options),
Tika Pipes (pipeline + Pipes-mode tuning),
Tika Server (REST + server config), or
Tika CLI (tika-app flags). The
Setting Limits page is the model — it
covers the Java API and the JSON config side by side. Filing issues against specific
advanced pages where the JSON/CLI equivalent isn’t documented yet helps us
prioritize the gap.
|
Running untrusted content safely
-
Robustness — process isolation and fault tolerance
-
Setting Limits — depth, size, timeout, and metadata bounds
-
TikaInputStream and Spooling — buffering, rewinding, and spilling to disk
Detection and text quality
-
Language Detection — the built-in CharSoup detector
-
Building the Language Detector — training and benchmarks
-
CharSoup Supported Languages — the 204 output classes
-
Text Quality Scoring (Junk Detection) — scoring text for corruption
-
Building the Junk Detector — training and model format
-
Charset Detection Pipeline — how the encoding-detector chain arbitrates
-
ZIP Detection and Salvaging — recovering truncated ZIP-based files
Parsing internals
-
Embedded Document Metadata — how Tika tracks embedded documents and their paths
-
Running a Local VLM Server — an OpenAI-compatible endpoint for air-gapped OCR