Support batch processing over the REST API. (#1806)

* Support batch processing over the REST API.

* Partially fix e2e tests

* Fix e2e tests

* ruff

* ruff

* consistent use of strings

* Update API docs

* Support batch processing over the REST API.

* Partially fix e2e tests

* Fix e2e tests

* ruff

* ruff

* consistent use of strings

* Update API docs

docs/api-docs/api-docs.yml

@@ -40,13 +40,20 @@ paths:
           content:
             application/json:
               schema:
-                description: "A list analysis results"
-                type: array
-                items:
-                  $ref: "#/components/schemas/RecognizerResultWithAnalysisExplanation"
+                oneOf:
+                  - description: "A list of analysis results (when input text is a single string)"
+                    type: array
+                    items:
+                      $ref: "#/components/schemas/RecognizerResultWithAnalysisExplanation"
+                  - description: "A list of analysis result lists (when input text is an array of strings)"
+                    type: array
+                    items:
+                      type: array
+                      items:
+                        $ref: "#/components/schemas/RecognizerResultWithAnalysisExplanation"
 
               examples:
-                Enhanced response:
+                Enhanced response (single text):
                   value:
                     [
                       { "entity_type": "PERSON", "start": 0, "end": 10, "score": 0.85,
@@ -66,7 +73,7 @@ paths:
                         }
                       }
                     ]
-                Lean response:
+                Lean response (single text):
                   value:
                     [
                       {
@@ -77,6 +84,17 @@ paths:
                         "start": 30
                       }
                     ]
+                Batch response (array of texts):
+                  value:
+                    [
+                      [
+                        { "entity_type": "PERSON", "start": 0, "end": 10, "score": 0.85, "analysis_explanation": null },
+                        { "entity_type": "US_DRIVER_LICENSE", "start": 30, "end": 38, "score": 0.6499999999999999, "analysis_explanation": null }
+                      ],
+                      [
+                        { "entity_type": "EMAIL_ADDRESS", "start": 15, "end": 35, "score": 0.95, "analysis_explanation": null }
+                      ]
+                    ]
 
   /recognizers:
     get:
@@ -272,13 +290,19 @@ components:
           schema:
             $ref: "#/components/schemas/AnalyzeRequest"
           examples:
-            Minimal Request:
+            Minimal Request (single text):
               value:
                 {
                   "text": "John Smith drivers license is AC432223",
                   "language": "en"
                 }
-            Enhanced Request :
+            Minimal Request (batch request):
+              value:
+                {
+                  "text": ["John Smith drivers license is AC432223", "John Smith drivers license is AC432223 and the zip code is 12345"],
+                  "language": "en"
+                }
+            Enhanced Request (single text):
               value:
                 {
                   "text": "John Smith drivers license is AC432223 and the zip code is 12345",
@@ -370,8 +394,12 @@ components:
         - language
       properties:
         text:
-          type: string
-          description: "The text to analyze"
+          oneOf:
+            - type: string
+            - type: array
+              items:
+                type: string
+          description: "The text to analyze. Can be a single string or an array of strings."
           example: "hello world, my name is Jane Doe. My number is: 034453334"
         language:
           type: string

e2e-tests/tests/test_api_analyzer.py

@@ -572,3 +572,34 @@ def test_given_regex_flags_and_normal_entities_are_returned():
     assert equal_json_strings(
         expected_response, response_content
     )
+
+@pytest.mark.api
+def test_given_a_correct_analyze_input_then_return_full_response_batched():
+    request_body = """
+    {
+        "text": [
+            "John Smith drivers license is AC432223",
+            "John Smith drivers license is AC432223"
+        ],
+        "language": "en"
+    }
+    """
+
+    response_status, response_content = analyze(request_body)
+
+    expected_response = """
+    [
+        [
+            {"entity_type": "PERSON", "start": 0, "end": 10, "score": 0.85, "analysis_explanation":null},
+            {"entity_type": "US_DRIVER_LICENSE", "start": 30, "end": 38, "score": 0.6499999999999999, "analysis_explanation":null}
+        ],
+        [
+            {"entity_type": "PERSON", "start": 0, "end": 10, "score": 0.85, "analysis_explanation":null},
+            {"entity_type": "US_DRIVER_LICENSE", "start": 30, "end": 38, "score": 0.6499999999999999, "analysis_explanation":null}
+        ]
+    ]
+    """
+    assert response_status == 200
+    assert equal_json_strings(
+        expected_response, response_content
+    )

presidio-analyzer/app.py

@@ -8,10 +8,17 @@ from pathlib import Path
 from typing import Tuple
 
 from flask import Flask, Response, jsonify, request
-from presidio_analyzer import AnalyzerEngine, AnalyzerEngineProvider, AnalyzerRequest
+from presidio_analyzer import (
+    AnalyzerEngine,
+    AnalyzerEngineProvider,
+    AnalyzerRequest,
+    BatchAnalyzerEngine,
+)
 from werkzeug.exceptions import HTTPException
 
 DEFAULT_PORT = "3000"
+DEFAULT_BATCH_SIZE = "500"
+DEFAULT_N_PROCESS = "1"
 
 LOGGING_CONF_FILE = "logging.ini"
 
@@ -46,6 +53,8 @@ class Server:
             nlp_engine_conf_file=nlp_engine_conf_file,
             recognizer_registry_conf_file=recognizer_registry_conf_file,
         ).create_engine()
+
+        self.batch_engine = BatchAnalyzerEngine(self.engine)
         self.logger.info(WELCOME_MESSAGE)
 
         @self.app.route("/health")
@@ -62,11 +71,21 @@ class Server:
                 if not req_data.text:
                     raise Exception("No text provided")
 
+                batch_request = isinstance(req_data.text, list)
+                batch = req_data.text if batch_request else [req_data.text]
+
                 if not req_data.language:
                     raise Exception("No language provided")
+                else:
+                    # Make sure the language is supported by the engine.
+                    self.engine.get_supported_entities(req_data.language)
 
-                recognizer_result_list = self.engine.analyze(
-                    text=req_data.text,
+                iterator = self.batch_engine.analyze_iterator(
+                    texts=batch,
+                    batch_size=min(
+                        len(batch),
+                        int(os.environ.get("BATCH_SIZE", DEFAULT_BATCH_SIZE))
+                    ),
                     language=req_data.language,
                     correlation_id=req_data.correlation_id,
                     score_threshold=req_data.score_threshold,
@@ -77,12 +96,19 @@ class Server:
                     allow_list=req_data.allow_list,
                     allow_list_match=req_data.allow_list_match,
                     regex_flags=req_data.regex_flags,
+                    n_process=min(
+                        len(batch),
+                        int(os.environ.get("N_PROCESS", DEFAULT_N_PROCESS))
+                    )
                 )
-                _exclude_attributes_from_dto(recognizer_result_list)
+                results = []
+                for recognizer_result_list in iterator:
+                    _exclude_attributes_from_dto(recognizer_result_list)
+                    results.append(recognizer_result_list)
 
                 return Response(
                     json.dumps(
-                        recognizer_result_list,
+                        results if batch_request else results[0],
                         default=lambda o: o.to_dict(),
                         sort_keys=True,
                     ),