{ "openapi": "3.0.1", "info": { "title": "PDF Text Extractor: OCR, Markdown, Page Checks", "description": "Read your PDFs and get the text plus a verdict on every page — the file's own layer, OCR for scans at no per-page charge, or a row saying why it couldn't be read. A wrong layer is flagged, not passed off as clean. Full text or Markdown, page ranges, locked files, uploads. Failed files cost nothing.", "version": "0.1", "x-build-id": "1J59vdqnft5igqczu" }, "servers": [ { "url": "https://api.apify.com/v2" } ], "paths": { "/acts/pradio~pdf-text-verifier/run-sync-get-dataset-items": { "post": { "operationId": "run-sync-get-dataset-items-pradio-pdf-text-verifier", "x-openai-isConsequential": false, "summary": "Executes an Actor, waits for its completion, and returns Actor's dataset items in response.", "tags": [ "Run Actor" ], "requestBody": { "required": true, "content": { "application/json": { "schema": { "$ref": "#/components/schemas/inputSchema" } } } }, "parameters": [ { "name": "token", "in": "query", "required": true, "schema": { "type": "string" }, "description": "Enter your Apify token here" } ], "responses": { "200": { "description": "OK" } } } }, "/acts/pradio~pdf-text-verifier/runs": { "post": { "operationId": "runs-sync-pradio-pdf-text-verifier", "x-openai-isConsequential": false, "summary": "Executes an Actor and returns information about the initiated run in response.", "tags": [ "Run Actor" ], "requestBody": { "required": true, "content": { "application/json": { "schema": { "$ref": "#/components/schemas/inputSchema" } } } }, "parameters": [ { "name": "token", "in": "query", "required": true, "schema": { "type": "string" }, "description": "Enter your Apify token here" } ], "responses": { "200": { "description": "OK", "content": { "application/json": { "schema": { "$ref": "#/components/schemas/runsResponseSchema" } } } } } } }, "/acts/pradio~pdf-text-verifier/run-sync": { "post": { "operationId": "run-sync-pradio-pdf-text-verifier", "x-openai-isConsequential": false, "summary": "Executes an Actor, waits for completion, and returns the OUTPUT from Key-value store in response.", "tags": [ "Run Actor" ], "requestBody": { "required": true, "content": { "application/json": { "schema": { "$ref": "#/components/schemas/inputSchema" } } } }, "parameters": [ { "name": "token", "in": "query", "required": true, "schema": { "type": "string" }, "description": "Enter your Apify token here" } ], "responses": { "200": { "description": "OK" } } } } }, "components": { "schemas": { "inputSchema": { "type": "object", "properties": { "urls": { "title": "PDF links", "type": "array", "description": "the PDF links, one per line — a comma or semicolon splits a pasted line. Drive and Dropbox share links work too. If the same file comes up twice we mark it a duplicate instead of reading it again; a page that only lists more links isn't followed", "items": { "type": "string" } }, "pdfFile": { "title": "One PDF file (upload or a web link)", "type": "string", "description": "pick the file in the Console's file picker, or pass its web URL through the API — a path on your machine isn't a URL. The same file twice is marked duplicate, not read again" }, "inputDatasetId": { "title": "A saved dataset of PDF links", "type": "string", "description": "the id of a saved Apify dataset — the letters and digits under its name in Storage — shared so anyone who has it can read it; it needs the id itself, not a link. Each item is read for the link in the url column, or whichever column urlField names; a duplicate is marked, not read again" }, "urlField": { "title": "Column each link is in", "type": "string", "description": "leave it as url unless your saved table names its link column something else; a row that is itself just a link is read as is", "default": "url" }, "outputFormat": { "title": "Output format", "enum": [ "text", "markdown", "both" ], "type": "string", "description": "what each row's whole-document columns carry: the text (default), Markdown with the layout's headings and lists, or both; per-page results carry the same text either way", "default": "text" }, "keepRows": { "title": "Which rows to keep", "enum": [ "all", "answered", "flagged", "missed" ], "type": "string", "description": "which results a run keeps: everything, only files that read, only flagged ones (suspect or unreadable), or only failures. A failure comes back as a free row naming it, and an entry your filter drops still shows up named — never billed", "default": "all" }, "cleanText": { "title": "Clean up the text", "type": "boolean", "description": "drop the headers and footers repeated on 30% or more of pages, page-number lines, space runs and split hyphens; the row lists what it removed, and per-page text stays as read", "default": false }, "extractTables": { "title": "Extract tables", "type": "boolean", "description": "rebuild the tables each page's layout carries — every table comes back with its page, rows of cells and a note on whether the rebuild is certain. A page read by OCR has no layout to work from; your text and page results stay as they are", "default": false }, "extractImages": { "title": "Extract embedded images", "type": "boolean", "description": "save each embedded image to the run's key-value store as a PNG, with its page, size and link listed in the row's images field; at most 200 a document", "default": false }, "skipEmptyPages": { "title": "Skip blank pages", "type": "boolean", "description": "leave a page carrying nothing — no text layer, no OCR words — out of the page results and the counts. A page the OCR cap stopped stays listed: never read isn't blank", "default": false }, "pageRange": { "title": "Page range", "type": "string", "description": "which pages to read in each file: 1-5 or 2-5,8, 10- to the end, -3 for the last three; empty reads everything. A typo names the entry in the results and bills nothing" }, "mustContain": { "title": "Must contain phrases", "type": "array", "description": "up to 20 phrases every file must carry, matched as written ignoring case; each row says found, which pages and the first matching line. More than 20, or an entry that isn't text, is refused", "items": { "type": "string" } }, "pdfPassword": { "title": "PDF password", "type": "string", "description": "the password that opens your protected PDFs — an entry's own password wins. A file that needs one and gets none comes back named and free" }, "ocrLanguage": { "title": "Language of scanned pages", "type": "string", "description": "the language your scanned pages are written in, English by default; join codes with + (eng+deu) for a document in two. A code the engine doesn't ship is refused before anything reads; OCR runs where a page has no text layer, and on every page when the page check is on", "default": "eng" }, "verifyPages": { "title": "Check each page's text against a scan (charged per page checked)", "type": "boolean", "description": "render every page and re-read the picture by OCR, flagging a born-digital text layer that disagrees. A scan's own transcript cannot be judged this way either: the fresh OCR is a second transcript, not a witness. It comes back verify_inconclusive, like a page whose read was too weak to judge. Inconclusive pages keep their text and bill nothing. Billed $0.008 per page the check verifies", "default": false }, "pdfTextMode": { "title": "Where a page's text comes from (leave Automatic)", "enum": [ "auto", "text", "ocr" ], "type": "string", "description": "auto reads the embedded layer and OCRs a page that has no usable one; text never runs OCR, so a layerless page answers no_layer with nothing; ocr reads every page by OCR and ships that text whatever the layer holds. With the page check on, text is refused", "default": "auto" }, "autoRotate": { "title": "Auto-rotate scans", "type": "boolean", "description": "retry a badly read OCR page turned 90, 180 and 270 degrees and keep the best read, for sideways scans; turned pages are marked in the page results", "default": true }, "dpi": { "title": "OCR image resolution (DPI)", "minimum": 72, "maximum": 400, "type": "integer", "description": "the resolution pages are rendered at for OCR, 72 to 400; empty applies 144 — the setting the stated rates were measured at" }, "preprocessMode": { "title": "Clean up scans before OCR (leave None)", "enum": [ "none", "grayscale", "contrast", "binarise" ], "type": "string", "description": "a cleanup the rendered page gets before OCR reads it: grayscale drops the colour, contrast stretches light and dark, binarise makes every pixel black or white. For faded or noisy scans; none leaves the page as rendered", "default": "none" }, "minOcrConfidence": { "title": "Minimum OCR confidence", "minimum": 0, "maximum": 100, "type": "integer", "description": "leave empty to flag nothing; 0 to 100 flags an OCR-read page whose confidence falls under it. The words stay, marked in the page results and the low-confidence list" }, "segmentationMode": { "title": "Page layout for scanned pages (leave Automatic)", "enum": [ "auto", "auto_osd", "single_column", "single_block", "single_line", "single_word", "sparse_text", "sparse_text_osd", "raw_line" ], "type": "string", "description": "leave Automatic unless a scan reads wrong: tells OCR whether the page is one column, a block, a line, a word or scattered text; the orientation choices also detect rotation", "default": "auto" }, "includeWords": { "title": "List each word OCR found", "type": "boolean", "description": "each OCR-read page lists every word it found with its confidence and position, for checking a scan word by word; a very large row may shorten the list and says so", "default": false }, "chunkSize": { "title": "Chunk size (characters)", "minimum": 200, "maximum": 20000, "type": "integer", "description": "split each document's text into chunks of about this size, each naming its pages; cuts fall at a paragraph, line or space. Empty for none; 200 to 20000" }, "chunkOverlap": { "title": "Characters repeated between chunks", "minimum": 0, "maximum": 10000, "type": "integer", "description": "characters repeated from a chunk's end into the next so a sentence cut at an edge stays whole; only used with Chunk size, at most half of it", "default": 200 }, "maxItems": { "title": "Maximum items", "minimum": 1, "maximum": 100000, "type": "integer", "description": "the most rows one run returns, free rows counting too. Entries past it are named on a free cap_reached row — or on the run's last row when the cap is 1", "default": 100 }, "maxPagesPerDocument": { "title": "Maximum pages per document", "minimum": 1, "maximum": 50000, "type": "integer", "description": "the most pages one file contributes; a longer one is read to here and marked pages_truncated. The cap also bounds a long file's extra-page charges", "default": 300 }, "ocrMaxPagesPerDocument": { "title": "Max pages to OCR per document", "minimum": 1, "maximum": 50000, "type": "integer", "description": "the most pages of one file OCR may read, never more than Maximum pages per document; leave empty for no limit. A page past the cap is marked, never skipped" }, "maxConcurrency": { "title": "Documents read at once", "minimum": 1, "maximum": 5, "type": "integer", "description": "how many files read at once, up to 5 (empty means 5). Rows keep your order and OCR still reads a page at a time; a run that renders (the page check, an OCR mode, image extraction) reads fewer at once, held by the memory the run was given — on a small allocation use Low-memory mode", "default": 5 }, "lowMemory": { "title": "Low-memory mode", "type": "boolean", "description": "read one file at a time so the buffers only ever hold one: slower, but the biggest allowed files fit 1024 MB. Ignores Documents read at once", "default": false }, "requestTimeoutSecs": { "title": "Fetch timeout per document (seconds)", "minimum": 5, "maximum": 600, "type": "integer", "description": "seconds one file's download may take before its row answers timeout, 5 to 600; empty keeps 60. A host asking for a short wait is waited out once" }, "maxDocumentSizeMb": { "title": "Maximum document size (MB)", "minimum": 1, "maximum": 250, "type": "integer", "description": "a file over this many megabytes isn't fetched — you get a free too_large row naming its size; empty keeps 80" }, "onlyNew": { "title": "Only new or changed documents", "type": "boolean", "description": "for lists you run again: skip a file an earlier run already returned when the file and your options match. Skipped entries are free and each named on its row; matching is by the file's fingerprint only", "default": false } } }, "runsResponseSchema": { "type": "object", "properties": { "data": { "type": "object", "properties": { "id": { "type": "string" }, "actId": { "type": "string" }, "userId": { "type": "string" }, "startedAt": { "type": "string", "format": "date-time", "example": "2025-01-08T00:00:00.000Z" }, "finishedAt": { "type": "string", "format": "date-time", "example": "2025-01-08T00:00:00.000Z" }, "status": { "type": "string", "example": "READY" }, "meta": { "type": "object", "properties": { "origin": { "type": "string", "example": "API" }, "userAgent": { "type": "string" } } }, "stats": { "type": "object", "properties": { "inputBodyLen": { "type": "integer", "example": 2000 }, "rebootCount": { "type": "integer", "example": 0 }, "restartCount": { "type": "integer", "example": 0 }, "resurrectCount": { "type": "integer", "example": 0 }, "computeUnits": { "type": "integer", "example": 0 } } }, "options": { "type": "object", "properties": { "build": { "type": "string", "example": "latest" }, "timeoutSecs": { "type": "integer", "example": 300 }, "memoryMbytes": { "type": "integer", "example": 1024 }, "diskMbytes": { "type": "integer", "example": 2048 } } }, "buildId": { "type": "string" }, "defaultKeyValueStoreId": { "type": "string" }, "defaultDatasetId": { "type": "string" }, "defaultRequestQueueId": { "type": "string" }, "buildNumber": { "type": "string", "example": "1.0.0" }, "containerUrl": { "type": "string" }, "usage": { "type": "object", "properties": { "ACTOR_COMPUTE_UNITS": { "type": "integer", "example": 0 }, "DATASET_READS": { "type": "integer", "example": 0 }, "DATASET_WRITES": { "type": "integer", "example": 0 }, "KEY_VALUE_STORE_READS": { "type": "integer", "example": 0 }, "KEY_VALUE_STORE_WRITES": { "type": "integer", "example": 1 }, "KEY_VALUE_STORE_LISTS": { "type": "integer", "example": 0 }, "REQUEST_QUEUE_READS": { "type": "integer", "example": 0 }, "REQUEST_QUEUE_WRITES": { "type": "integer", "example": 0 }, "DATA_TRANSFER_INTERNAL_GBYTES": { "type": "integer", "example": 0 }, "DATA_TRANSFER_EXTERNAL_GBYTES": { "type": "integer", "example": 0 }, "PROXY_RESIDENTIAL_TRANSFER_GBYTES": { "type": "integer", "example": 0 }, "PROXY_SERPS": { "type": "integer", "example": 0 }, "PROXY_UNBLOCKER_UNITS": { "type": "integer", "example": 0 } } }, "usageTotalUsd": { "type": "number", "example": 0.00005 }, "usageUsd": { "type": "object", "properties": { "ACTOR_COMPUTE_UNITS": { "type": "integer", "example": 0 }, "DATASET_READS": { "type": "integer", "example": 0 }, "DATASET_WRITES": { "type": "integer", "example": 0 }, "KEY_VALUE_STORE_READS": { "type": "integer", "example": 0 }, "KEY_VALUE_STORE_WRITES": { "type": "number", "example": 0.00005 }, "KEY_VALUE_STORE_LISTS": { "type": "integer", "example": 0 }, "REQUEST_QUEUE_READS": { "type": "integer", "example": 0 }, "REQUEST_QUEUE_WRITES": { "type": "integer", "example": 0 }, "DATA_TRANSFER_INTERNAL_GBYTES": { "type": "integer", "example": 0 }, "DATA_TRANSFER_EXTERNAL_GBYTES": { "type": "integer", "example": 0 }, "PROXY_RESIDENTIAL_TRANSFER_GBYTES": { "type": "integer", "example": 0 }, "PROXY_SERPS": { "type": "integer", "example": 0 }, "PROXY_UNBLOCKER_UNITS": { "type": "integer", "example": 0 } } } } } } } } } }