From 75f9f527954aac2b84672055bdb0447eea0c3283 Mon Sep 17 00:00:00 2001 From: mahour Date: Wed, 20 Nov 2024 19:09:28 +0530 Subject: [PATCH 1/2] updated extract schema --- src/pages/resources/extractJSONOutputSchema2.json | 4 ++++ .../resources/extractJSONOutputSchemaStylingInfo.json | 7 ++++++- static/extractJSONOutputSchema2.json | 4 ++++ static/extractJSONOutputSchemaStylingInfo.json | 7 ++++++- 4 files changed, 20 insertions(+), 2 deletions(-) diff --git a/src/pages/resources/extractJSONOutputSchema2.json b/src/pages/resources/extractJSONOutputSchema2.json index e10276334..63ddc93fb 100644 --- a/src/pages/resources/extractJSONOutputSchema2.json +++ b/src/pages/resources/extractJSONOutputSchema2.json @@ -700,6 +700,10 @@ "items": { "type": "string" } + }, + "alternate_text": { + "description": "Provides alternate text present in the input pdf structure mapped to the updated structure", + "type": "string" } }, "required": [ diff --git a/src/pages/resources/extractJSONOutputSchemaStylingInfo.json b/src/pages/resources/extractJSONOutputSchemaStylingInfo.json index 7b1283fe0..ec6292d47 100644 --- a/src/pages/resources/extractJSONOutputSchemaStylingInfo.json +++ b/src/pages/resources/extractJSONOutputSchemaStylingInfo.json @@ -580,6 +580,10 @@ "minItems": 1, "items": { "type": "string" + }, + "alternate_text": { + "description": "Provides alternate text present in the input pdf structure mapped to the updated structure", + "type": "string" } }, "Kids": { @@ -620,7 +624,8 @@ "Image", "Reference", "Kids", - "filePaths" + "filePaths", + "alternate_text" ] } }, diff --git a/static/extractJSONOutputSchema2.json b/static/extractJSONOutputSchema2.json index e10276334..63ddc93fb 100644 --- a/static/extractJSONOutputSchema2.json +++ b/static/extractJSONOutputSchema2.json @@ -700,6 +700,10 @@ "items": { "type": "string" } + }, + "alternate_text": { + "description": "Provides alternate text present in the input pdf structure mapped to the updated structure", + "type": "string" } }, "required": [ diff --git a/static/extractJSONOutputSchemaStylingInfo.json b/static/extractJSONOutputSchemaStylingInfo.json index 7b1283fe0..ec6292d47 100644 --- a/static/extractJSONOutputSchemaStylingInfo.json +++ b/static/extractJSONOutputSchemaStylingInfo.json @@ -580,6 +580,10 @@ "minItems": 1, "items": { "type": "string" + }, + "alternate_text": { + "description": "Provides alternate text present in the input pdf structure mapped to the updated structure", + "type": "string" } }, "Kids": { @@ -620,7 +624,8 @@ "Image", "Reference", "Kids", - "filePaths" + "filePaths", + "alternate_text" ] } }, From abe6ab2f241566c7c5da195ab1ab762e7a5bbc6e Mon Sep 17 00:00:00 2001 From: mahour Date: Fri, 22 Nov 2024 17:01:45 +0530 Subject: [PATCH 2/2] updated extract schema --- .../resources/extractJSONOutputSchema2.json | 9 +++++++-- .../extractJSONOutputSchemaStylingInfo.json | 18 +++++++++++++----- static/extractJSONOutputSchema2.json | 9 +++++++-- static/extractJSONOutputSchemaStylingInfo.json | 18 +++++++++++++----- 4 files changed, 40 insertions(+), 14 deletions(-) diff --git a/src/pages/resources/extractJSONOutputSchema2.json b/src/pages/resources/extractJSONOutputSchema2.json index 63ddc93fb..e35261d4f 100644 --- a/src/pages/resources/extractJSONOutputSchema2.json +++ b/src/pages/resources/extractJSONOutputSchema2.json @@ -85,7 +85,7 @@ "description": "Document contains AcroForms or not", "type": "boolean" }, - "is_signed": { + "is_digitally_signed": { "description": "Document is digitally signed or not", "type": "boolean" }, @@ -127,7 +127,7 @@ "has_embedded_files", "is_XFA", "is_certified", - "is_signed", + "is_digitally_signed", "is_encrypted", "language", "page_count", @@ -224,6 +224,11 @@ "type": "boolean", "default": false }, + "ObjectID": { + "description": "Unique identifier for the element", + "type": "integer", + "minimum": 0 + }, "Lang": { "description": "BCP-47 code for language of text elements", "type": "string" diff --git a/src/pages/resources/extractJSONOutputSchemaStylingInfo.json b/src/pages/resources/extractJSONOutputSchemaStylingInfo.json index ec6292d47..e12ffc28c 100644 --- a/src/pages/resources/extractJSONOutputSchemaStylingInfo.json +++ b/src/pages/resources/extractJSONOutputSchemaStylingInfo.json @@ -111,6 +111,11 @@ "type": "boolean", "default": false }, + "ObjectID": { + "description": "Unique identifier for the element", + "type": "integer", + "minimum": 0 + }, "Lang": { "description": "BCP-47 code for language of text elements", "type": "string" @@ -582,7 +587,7 @@ "type": "string" }, "alternate_text": { - "description": "Provides alternate text present in the input pdf structure mapped to the updated structure", + "description": "Provides alternate text present in the input pdf structure use to mapped to the updated structure", "type": "string" } }, @@ -625,7 +630,8 @@ "Reference", "Kids", "filePaths", - "alternate_text" + "alternate_text", + "ObjectID" ] } }, @@ -647,7 +653,9 @@ "Rotation", "Image", "Reference", - "filePaths" + "filePaths", + "alternate_text", + "ObjectID" ] } } @@ -710,7 +718,7 @@ "description": "Document contains AcroForms or not", "type": "boolean" }, - "is_signed": { + " is_digitally_signed": { "description": "Document is digitally signed or not", "type": "boolean" }, @@ -752,7 +760,7 @@ "has_embedded_files", "is_XFA", "is_certified", - "is_signed", + "is_digitally_signed", "is_encrypted", "language", "page_count", diff --git a/static/extractJSONOutputSchema2.json b/static/extractJSONOutputSchema2.json index 63ddc93fb..e35261d4f 100644 --- a/static/extractJSONOutputSchema2.json +++ b/static/extractJSONOutputSchema2.json @@ -85,7 +85,7 @@ "description": "Document contains AcroForms or not", "type": "boolean" }, - "is_signed": { + "is_digitally_signed": { "description": "Document is digitally signed or not", "type": "boolean" }, @@ -127,7 +127,7 @@ "has_embedded_files", "is_XFA", "is_certified", - "is_signed", + "is_digitally_signed", "is_encrypted", "language", "page_count", @@ -224,6 +224,11 @@ "type": "boolean", "default": false }, + "ObjectID": { + "description": "Unique identifier for the element", + "type": "integer", + "minimum": 0 + }, "Lang": { "description": "BCP-47 code for language of text elements", "type": "string" diff --git a/static/extractJSONOutputSchemaStylingInfo.json b/static/extractJSONOutputSchemaStylingInfo.json index ec6292d47..e12ffc28c 100644 --- a/static/extractJSONOutputSchemaStylingInfo.json +++ b/static/extractJSONOutputSchemaStylingInfo.json @@ -111,6 +111,11 @@ "type": "boolean", "default": false }, + "ObjectID": { + "description": "Unique identifier for the element", + "type": "integer", + "minimum": 0 + }, "Lang": { "description": "BCP-47 code for language of text elements", "type": "string" @@ -582,7 +587,7 @@ "type": "string" }, "alternate_text": { - "description": "Provides alternate text present in the input pdf structure mapped to the updated structure", + "description": "Provides alternate text present in the input pdf structure use to mapped to the updated structure", "type": "string" } }, @@ -625,7 +630,8 @@ "Reference", "Kids", "filePaths", - "alternate_text" + "alternate_text", + "ObjectID" ] } }, @@ -647,7 +653,9 @@ "Rotation", "Image", "Reference", - "filePaths" + "filePaths", + "alternate_text", + "ObjectID" ] } } @@ -710,7 +718,7 @@ "description": "Document contains AcroForms or not", "type": "boolean" }, - "is_signed": { + " is_digitally_signed": { "description": "Document is digitally signed or not", "type": "boolean" }, @@ -752,7 +760,7 @@ "has_embedded_files", "is_XFA", "is_certified", - "is_signed", + "is_digitally_signed", "is_encrypted", "language", "page_count",