Skip to content

Commit f9085ee

Browse files
datalogics-cgreendatalogics-kam
datalogics-cgreen
authored and
datalogics-kam
committed
Add Convert to PDF
- Add methods for conversion from file and conversion from url - Do argument validation based on input file format Assisted-by: Codex
1 parent 0ab9fdd commit f9085ee

4 files changed

Lines changed: 422 additions & 0 deletions

File tree

‎src/pdfrest/client.py‎

Lines changed: 196 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -76,6 +76,8 @@
7676
BasePdfRestGraphicPayload,
7777
BmpPdfRestPayload,
7878
ConvertToMarkdownPayload,
79+
ConvertToPdfPayload,
80+
ConvertUrlsToPdfPayload,
7981
DeletePayload,
8082
ExtractImagesPayload,
8183
ExtractTextPayload,
@@ -121,10 +123,16 @@
121123
FlattenQuality,
122124
GifColorModel,
123125
GraphicSmoothing,
126+
HtmlPageOrientation,
127+
HtmlPageSize,
128+
HtmlWebLayout,
124129
JpegColorModel,
125130
OcrLanguage,
126131
PdfAddTextObject,
127132
PdfAType,
133+
PdfConversionCompression,
134+
PdfConversionDownsample,
135+
PdfConversionLocale,
128136
PdfInfoQuery,
129137
PdfMergeInput,
130138
PdfPageSelection,
@@ -3175,6 +3183,100 @@ def rasterize_pdf(
31753183
timeout=timeout,
31763184
)
31773185

3186+
def convert_to_pdf(
3187+
self,
3188+
file: PdfRestFile | Sequence[PdfRestFile],
3189+
*,
3190+
output: str | None = None,
3191+
compression: PdfConversionCompression | None = None,
3192+
downsample: PdfConversionDownsample | None = None,
3193+
tagged_pdf: bool | None = None,
3194+
locale: PdfConversionLocale | None = None,
3195+
page_size: HtmlPageSize | None = None,
3196+
page_margin: str | None = None,
3197+
page_orientation: HtmlPageOrientation | None = None,
3198+
web_layout: HtmlWebLayout | None = None,
3199+
extra_query: Query | None = None,
3200+
extra_headers: AnyMapping | None = None,
3201+
extra_body: Body | None = None,
3202+
timeout: TimeoutTypes | None = None,
3203+
) -> PdfRestFileBasedResponse:
3204+
"""Convert a supported file type to PDF."""
3205+
3206+
payload: dict[str, Any] = {"files": file}
3207+
if output is not None:
3208+
payload["output"] = output
3209+
if compression is not None:
3210+
payload["compression"] = compression
3211+
if downsample is not None:
3212+
payload["downsample"] = downsample
3213+
if tagged_pdf is not None:
3214+
payload["tagged_pdf"] = tagged_pdf
3215+
if locale is not None:
3216+
payload["locale"] = locale
3217+
if page_size is not None:
3218+
payload["page_size"] = page_size
3219+
if page_margin is not None:
3220+
payload["page_margin"] = page_margin
3221+
if page_orientation is not None:
3222+
payload["page_orientation"] = page_orientation
3223+
if web_layout is not None:
3224+
payload["web_layout"] = web_layout
3225+
3226+
return self._post_file_operation(
3227+
endpoint="/pdf",
3228+
payload=payload,
3229+
payload_model=ConvertToPdfPayload,
3230+
extra_query=extra_query,
3231+
extra_headers=extra_headers,
3232+
extra_body=extra_body,
3233+
timeout=timeout,
3234+
)
3235+
3236+
def convert_urls_to_pdf(
3237+
self,
3238+
urls: UrlInput,
3239+
*,
3240+
output: str | None = None,
3241+
compression: PdfConversionCompression | None = None,
3242+
downsample: PdfConversionDownsample | None = None,
3243+
page_size: HtmlPageSize | None = None,
3244+
page_margin: str | None = None,
3245+
page_orientation: HtmlPageOrientation | None = None,
3246+
web_layout: HtmlWebLayout | None = None,
3247+
extra_query: Query | None = None,
3248+
extra_headers: AnyMapping | None = None,
3249+
extra_body: Body | None = None,
3250+
timeout: TimeoutTypes | None = None,
3251+
) -> PdfRestFileBasedResponse:
3252+
"""Convert HTML content from one or more URLs to PDF."""
3253+
3254+
payload: dict[str, Any] = {"url": urls}
3255+
if output is not None:
3256+
payload["output"] = output
3257+
if compression is not None:
3258+
payload["compression"] = compression
3259+
if downsample is not None:
3260+
payload["downsample"] = downsample
3261+
if page_size is not None:
3262+
payload["page_size"] = page_size
3263+
if page_margin is not None:
3264+
payload["page_margin"] = page_margin
3265+
if page_orientation is not None:
3266+
payload["page_orientation"] = page_orientation
3267+
if web_layout is not None:
3268+
payload["web_layout"] = web_layout
3269+
3270+
return self._post_file_operation(
3271+
endpoint="/pdf",
3272+
payload=payload,
3273+
payload_model=ConvertUrlsToPdfPayload,
3274+
extra_query=extra_query,
3275+
extra_headers=extra_headers,
3276+
extra_body=extra_body,
3277+
timeout=timeout,
3278+
)
3279+
31783280
def convert_to_pdfa(
31793281
self,
31803282
file: PdfRestFile | Sequence[PdfRestFile],
@@ -4555,6 +4657,100 @@ async def rasterize_pdf(
45554657
timeout=timeout,
45564658
)
45574659

4660+
async def convert_to_pdf(
4661+
self,
4662+
file: PdfRestFile | Sequence[PdfRestFile],
4663+
*,
4664+
output: str | None = None,
4665+
compression: PdfConversionCompression | None = None,
4666+
downsample: PdfConversionDownsample | None = None,
4667+
tagged_pdf: bool | None = None,
4668+
locale: PdfConversionLocale | None = None,
4669+
page_size: HtmlPageSize | None = None,
4670+
page_margin: str | None = None,
4671+
page_orientation: HtmlPageOrientation | None = None,
4672+
web_layout: HtmlWebLayout | None = None,
4673+
extra_query: Query | None = None,
4674+
extra_headers: AnyMapping | None = None,
4675+
extra_body: Body | None = None,
4676+
timeout: TimeoutTypes | None = None,
4677+
) -> PdfRestFileBasedResponse:
4678+
"""Asynchronously convert a supported file type to PDF."""
4679+
4680+
payload: dict[str, Any] = {"files": file}
4681+
if output is not None:
4682+
payload["output"] = output
4683+
if compression is not None:
4684+
payload["compression"] = compression
4685+
if downsample is not None:
4686+
payload["downsample"] = downsample
4687+
if tagged_pdf is not None:
4688+
payload["tagged_pdf"] = tagged_pdf
4689+
if locale is not None:
4690+
payload["locale"] = locale
4691+
if page_size is not None:
4692+
payload["page_size"] = page_size
4693+
if page_margin is not None:
4694+
payload["page_margin"] = page_margin
4695+
if page_orientation is not None:
4696+
payload["page_orientation"] = page_orientation
4697+
if web_layout is not None:
4698+
payload["web_layout"] = web_layout
4699+
4700+
return await self._post_file_operation(
4701+
endpoint="/pdf",
4702+
payload=payload,
4703+
payload_model=ConvertToPdfPayload,
4704+
extra_query=extra_query,
4705+
extra_headers=extra_headers,
4706+
extra_body=extra_body,
4707+
timeout=timeout,
4708+
)
4709+
4710+
async def convert_urls_to_pdf(
4711+
self,
4712+
urls: UrlInput,
4713+
*,
4714+
output: str | None = None,
4715+
compression: PdfConversionCompression | None = None,
4716+
downsample: PdfConversionDownsample | None = None,
4717+
page_size: HtmlPageSize | None = None,
4718+
page_margin: str | None = None,
4719+
page_orientation: HtmlPageOrientation | None = None,
4720+
web_layout: HtmlWebLayout | None = None,
4721+
extra_query: Query | None = None,
4722+
extra_headers: AnyMapping | None = None,
4723+
extra_body: Body | None = None,
4724+
timeout: TimeoutTypes | None = None,
4725+
) -> PdfRestFileBasedResponse:
4726+
"""Asynchronously convert HTML content from one or more URLs to PDF."""
4727+
4728+
payload: dict[str, Any] = {"url": urls}
4729+
if output is not None:
4730+
payload["output"] = output
4731+
if compression is not None:
4732+
payload["compression"] = compression
4733+
if downsample is not None:
4734+
payload["downsample"] = downsample
4735+
if page_size is not None:
4736+
payload["page_size"] = page_size
4737+
if page_margin is not None:
4738+
payload["page_margin"] = page_margin
4739+
if page_orientation is not None:
4740+
payload["page_orientation"] = page_orientation
4741+
if web_layout is not None:
4742+
payload["web_layout"] = web_layout
4743+
4744+
return await self._post_file_operation(
4745+
endpoint="/pdf",
4746+
payload=payload,
4747+
payload_model=ConvertUrlsToPdfPayload,
4748+
extra_query=extra_query,
4749+
extra_headers=extra_headers,
4750+
extra_body=extra_body,
4751+
timeout=timeout,
4752+
)
4753+
45584754
async def convert_to_pdfa(
45594755
self,
45604756
file: PdfRestFile | Sequence[PdfRestFile],

0 commit comments

Comments
 (0)