While using the ChartFox interface for CYOW I found that the charts were not displayed (Cannot open document: ...) and traced this issue to the rasterizer assuming that charts are in PDF format.
This fix retrieves the content type header from the http transaction and passes that to the rasterizer (MuPDF/fitz) so that it is able to handle the data blob correctly (previously the blob type was hard-coded to application/pdf).
A newer version of the curl submodule was required in order to acess the header fields to get the content type. Since there is a dependency on the mbedtls module, and updating this turned out to be non-trivial, I have selected the latest version of the curl module that does not require any changes to the mbedtls module.
Since the purpose of this PR is to support downloaded charts that are not PDF formatted, I have renamed classes and files that had 'PDF' in their names to the more general 'Document', and similarly variables are renamed to reduce the possibility of confusion. I have also removed 2 files that are not linked into the build and appear to be redundant.
One of the settings in the json preferences file was called pdfreading. Since this applies to all documents, it has been renamed to docreading, and code is added to support upgrading of the preferences file on first run of Avitab after this PR has been included in the build.
The original PDFSource class, now renamed as DocumentSource overloads downloaded documents and local file documents, with some parts of the class only relevant to local files. The final refactoring update was to create subclasses of DocumentSource to represent LocalFileSource and DownloadedSource documents.
The actual functional change that obtains the content type of the downloaded document and passes this to the rasterizer is in the final commit and is relatively trivial!
This is a revival of https://github.com/fpw/avitab/pull/207, since I was unable to reopen that request after I closed it (temporarily I had hoped!)
While using the ChartFox interface for CYOW I found that the charts were not displayed (Cannot open document: ...) and traced this issue to the rasterizer assuming that charts are in PDF format.
This fix retrieves the content type header from the http transaction and passes that to the rasterizer (MuPDF/fitz) so that it is able to handle the data blob correctly (previously the blob type was hard-coded to application/pdf).
A newer version of the curl submodule was required in order to acess the header fields to get the content type. Since there is a dependency on the mbedtls module, and updating this turned out to be non-trivial, I have selected the latest version of the curl module that does not require any changes to the mbedtls module.
Since the purpose of this PR is to support downloaded charts that are not PDF formatted, I have renamed classes and files that had 'PDF' in their names to the more general 'Document', and similarly variables are renamed to reduce the possibility of confusion. I have also removed 2 files that are not linked into the build and appear to be redundant.
One of the settings in the json preferences file was called pdfreading. Since this applies to all documents, it has been renamed to docreading, and code is added to support upgrading of the preferences file on first run of Avitab after this PR has been included in the build.
The original PDFSource class, now renamed as DocumentSource overloads downloaded documents and local file documents, with some parts of the class only relevant to local files. The final refactoring update was to create subclasses of DocumentSource to represent LocalFileSource and DownloadedSource documents.
The actual functional change that obtains the content type of the downloaded document and passes this to the rasterizer is in the final commit and is relatively trivial!