diff options
Diffstat (limited to 'scripts/context/lua/mtx-pdf.lua')
| -rw-r--r-- | scripts/context/lua/mtx-pdf.lua | 288 |
1 files changed, 189 insertions, 99 deletions
diff --git a/scripts/context/lua/mtx-pdf.lua b/scripts/context/lua/mtx-pdf.lua index 371b7f319..2e73fa841 100644 --- a/scripts/context/lua/mtx-pdf.lua +++ b/scripts/context/lua/mtx-pdf.lua @@ -7,7 +7,7 @@ if not modules then modules = { } end modules ['mtx-pdf'] = { } local tonumber = tonumber -local format, gmatch = string.format, string.gmatch +local format, gmatch, gsub = string.format, string.gmatch, string.gsub local utfchar = utf.char local concat = table.concat local setmetatableindex, sortedhash, sortedkeys = table.setmetatableindex, table.sortedhash, table.sortedkeys @@ -25,8 +25,13 @@ local helpinfo = [[ <subcategory> <flag name="info"><short>show some info about the given file</short></flag> <flag name="metadata"><short>show metadata xml blob</short></flag> + <flag name="pretty"><short>replace newlines in metadata</short></flag> <flag name="fonts"><short>show used fonts (<ref name="detail)"/></short></flag> - <flag name="linearize"><short>linearize given file</short></flag> + </subcategory> + <subcategory> + <example><command>mtxrun --script pdf --info foo.pdf</command></example> + <example><command>mtxrun --script pdf --metadata foo.pdf</command></example> + <example><command>mtxrun --script pdf --metadata --pretty foo.pdf</command></example> </subcategory> </category> </flags> @@ -41,11 +46,17 @@ local application = logs.application { local report = application.report -dofile(resolvers.findfile("lpdf-epd.lua","tex")) +if pdfe then + dofile(resolvers.findfile("lpdf-pde.lua","tex")) +else + dofile(resolvers.findfile("lpdf-epd.lua","tex")) +end scripts = scripts or { } scripts.pdf = scripts.pdf or { } +local details = environment.argument("detail") or environment.argument("details") + local function loadpdffile(filename) if not filename or filename == "" then report("no filename given") @@ -64,60 +75,134 @@ end function scripts.pdf.info(filename) local pdffile = loadpdffile(filename) if pdffile then - local catalog = pdffile.Catalog - local info = pdffile.Info - local pages = pdffile.pages - local nofpages = pages.n -- no # yet. will be in 5.2 + local catalog = pdffile.Catalog + local info = pdffile.Info + local pages = pdffile.pages + local nofpages = pdffile.nofpages + + local unset = "<unset>" - report("filename > %s",filename) - report("pdf version > %s",catalog.Version) - report("number of pages > %s",nofpages) - report("title > %s",info.Title) - report("creator > %s",info.Creator) - report("producer > %s",info.Producer) - report("creation date > %s",info.CreationDate) - report("modification date > %s",info.ModDate) + report("%-17s > %s","filename", filename) + report("%-17s > %s","pdf version", catalog.Version or unset) + report("%-17s > %s","major version", pdffile.majorversion or unset) + report("%-17s > %s","minor version", pdffile.minorversion or unset) + report("%-17s > %s","number of pages", nofpages or 0) + report("%-17s > %s","title", info.Title or unset) + report("%-17s > %s","creator", info.Creator or unset) + report("%-17s > %s","producer", info.Producer or unset) + report("%-17s > %s","creation date", info.CreationDate or unset) + report("%-17s > %s","modification date", info.ModDate or unset) - local width, height, start - for i=1, nofpages do - local page = pages[i] - local bbox = page.CropBox or page.MediaBox - local w, h = bbox[4]-bbox[2],bbox[3]-bbox[1] - if w ~= width or h ~= height then - if start then - report("cropbox > pages: %s-%s, width: %s, height: %s",start,i-1,width,height) + local function somebox(what) + local box = string.lower(what) + local width, height, start + for i=1, nofpages do + local page = pages[i] + local bbox = page[what] or page.MediaBox or { 0, 0, 0, 0 } + local w, h = bbox[4]-bbox[2],bbox[3]-bbox[1] + if w ~= width or h ~= height then + if start then + report("%-17s > pages: %s-%s, width: %s, height: %s",box,start,i-1,width,height) + end + width, height, start = w, h, i end - width, height, start = w, h, i end + report("%-17s > pages: %s-%s, width: %s, height: %s",box,start,nofpages,width,height) end - report("cropbox > pages: %s-%s, width: %s, height: %s",start,nofpages,width,height) + + if details then + somebox("MediaBox") + somebox("ArtBox") + somebox("BleedBox") + somebox("CropBox") + somebox("TrimBox") + else + somebox("CropBox") + end + + -- if details then + local annotations = 0 + for i=1, nofpages do + local page = pages[i] + local a = page.Annots + if a then + annotations = annotations + #a + end + end + if annotations > 0 then + report("%-17s > %s", "annotations",annotations) + end + -- end + + -- if details then + local d = pdffile.destinations + local k = d and sortedkeys(d) + if k and #k > 0 then + report("%-17s > %s", "destinations",#k) + end + local d = pdffile.javascripts + local k = d and sortedkeys(d) + if k and #k > 0 then + report("%-17s > %s", "javascripts",#k) + end + local d = pdffile.widgets + if d and #d > 0 then + report("%-17s > %s", "widgets",#d) + end + local d = pdffile.embeddedfiles + local k = d and sortedkeys(d) + if k and #k > 0 then + report("%-17s > %s", "embeddedfiles",#k) + end + -- end + end end -function scripts.pdf.metadata(filename) +function scripts.pdf.metadata(filename,pretty) local pdffile = loadpdffile(filename) if pdffile then local catalog = pdffile.Catalog local metadata = catalog.Metadata if metadata then - report("metadata > \n\n%s\n",metadata()) + metadata = metadata() + if pretty then + metadata = gsub(metadata,"\r","\n") + end + report("metadata > \n\n%s\n",metadata) else report("no metadata") end end end +local expanded = lpdf.epdf.expanded + local function getfonts(pdffile) local usedfonts = { } - for i=1,pdffile.pages.n do - local page = pdffile.pages[i] - local fontlist = page.Resources.Font - if fontlist then - for k, v in next, lpdf.epdf.expand(fontlist) do - usedfonts[k] = lpdf.epdf.expand(v) + + local function collect(where,tag) + local resources = where.Resources + if resources then + local fontlist = resources.Font + if fontlist then + for k, v in expanded(fontlist) do + usedfonts[tag and (tag .. "." .. k) or k] = v + end + end + local objects = resources.XObject + if objects then + for k, v in expanded(objects) do + collect(v,tag and (tag .. "." .. k) or k) + end end end end + + for i=1,pdffile.nofpages do + collect(pdffile.pages[i]) + end + return usedfonts end @@ -125,7 +210,8 @@ local function getunicodes(font) local cid = font.ToUnicode if cid then cid = cid() - local counts = { } + local counts = { } + local indices = { } -- for s in gmatch(cid,"begincodespacerange%s*(.-)%s*endcodespacerange") do -- for a, b in gmatch(s,"<([^>]+)>%s+<([^>]+)>") do -- print(a,b) @@ -141,18 +227,20 @@ local function getunicodes(font) for i=first,last do local c = i + offset counts[c] = counts[c] + 1 + indices[i] = true end end end for s in gmatch(cid,"beginbfchar%s*(.-)%s*endbfchar") do for old, new in gmatch(s,"<([^>]+)>%s+<([^>]+)>") do + indices[tonumber(old,16)] = true for n in gmatch(new,"....") do local c = tonumber(n,16) counts[c] = counts[c] + 1 end end end - return counts + return counts, indices end end @@ -161,49 +249,93 @@ function scripts.pdf.fonts(filename) if pdffile then local usedfonts = getfonts(pdffile) local found = { } + local common = table.setmetatableindex("table") for k, v in table.sortedhash(usedfonts) do - local counts = getunicodes(v) - local codes = { } - local chars = { } - local freqs = { } + local basefont = v.BaseFont + local encoding = v.Encoding + local subtype = v.Subtype + local unicode = v.ToUnicode + local counts, + indices = getunicodes(v) + local codes = { } + local chars = { } + local freqs = { } + local names = { } if counts then codes = sortedkeys(counts) for i=1,#codes do local k = codes[i] - local c = utfchar(k) - chars[i] = c - freqs[i] = format("U+%05X %s %s",k,counts[k] > 1 and "+" or " ", c) + if k > 32 then + local c = utfchar(k) + chars[i] = c + freqs[i] = format("U+%05X %s %s",k,counts[k] > 1 and "+" or " ", c) + else + freqs[i] = format("U+%05X %s --",k,counts[k] > 1 and "+" or " ") + end + end + if basefont and unicode then + local b = gsub(basefont,"^.*%+","") + local c = common[b] + for k in next, indices do + c[k] = true + end end for i=1,#codes do codes[i] = format("U+%05X",codes[i]) end end + local d = encoding and encoding.Differences + if d then + for i=1,#d do + local di = d[i] + if type(di) == "string" then + names[#names+1] = di + end + end + end found[k] = { - basefont = v.BaseFont or "no basefont", - encoding = v.Encoding or "no encoding", - subtype = v.Subtype or "no subtype", - unicode = v.ToUnicode and "unicode" or "no unicode", + basefont = basefont or "no basefont", + encoding = (d and "custom n=" .. #d) or "no encoding", + subtype = subtype or "no subtype", + unicode = tounicode and "unicode" or "no vector", chars = chars, codes = codes, freqs = freqs, + names = names, } end - if environment.argument("detail") then + if details then for k, v in sortedhash(found) do - report("id : %s",k) - report("basefont : %s",v.basefont) - report("encoding : %s",v.encoding) - report("subtype : %s",v.subtype) - report("unicode : %s",v.unicode) - report("characters : %s", concat(v.chars," ")) - report("codepoints : %s", concat(v.codes," ")) + report("id : %s", k) + report("basefont : %s", v.basefont) + report("encoding : % t", v.names) + report("subtype : %s", v.subtype) + report("unicode : %s", v.unicode) + if #v.chars > 0 then + report("characters : % t", v.chars) + end + if #v.codes > 0 then + report("codepoints : % t", v.codes) + end + report("") + end + for k, v in sortedhash(common) do + report("basefont : %s",k) + report("indices : % t", sortedkeys(v)) report("") end else - local results = { { "id", "basefont", "encoding", "subtype", "unicode", "characters" } } + local haschar = false for k, v in sortedhash(found) do - results[#results+1] = { k, v.basefont, v.encoding, v.subtype, v.unicode, concat(v.chars," ") } + if #v.chars > 0 then + haschar = true + break + end + end + local results = { { "id", "basefont", "encoding", "subtype", "unicode", haschar and "characters" or nil } } + for k, v in sortedhash(found) do + results[#results+1] = { k, v.basefont, v.encoding, v.subtype, v.unicode, haschar and concat(v.chars," ") or nil } end utilities.formatters.formatcolumns(results) report(results[1]) @@ -216,46 +348,6 @@ function scripts.pdf.fonts(filename) end end --- this is a quick hack ... proof of concept .. will change (derived from luigi's example) ... --- i will make a ctx wrapper - -local qpdf -- just call qpdf, no need for a lib here - -function scripts.pdf.linearize(filename) - qpdf = qpdf or swiglib("qpdf.core") - local oldfile = filename or environment.files[1] - if not oldfile then - return - end - file.addsuffix(oldfile,"pdf") - if not lfs.isfile(oldfile) then - return - end - local newfile = environment.files[2] - if not newfile or file.removesuffix(oldfile) == file.removesuffix(newfile)then - newfile = file.addsuffix(file.removesuffix(oldfile) .. "-linearized","pdf") - end - local password = environment.arguments.password - local instance = qpdf.qpdf_init() - if bit32.band(qpdf.qpdf_read(instance,oldfile,password),qpdf.QPDF_ERRORS) ~= 0 then - report("unable to open input file") - elseif bit32.band(qpdf.qpdf_init_write(instance,newfile),qpdf.QPDF_ERRORS) ~= 0 then - report("unable to open output file") - else - report("linearizing %a into %a",oldfile,newfile) - qpdf.qpdf_set_static_ID(instance,qpdf.QPDF_TRUE) - qpdf.qpdf_set_linearization(instance,qpdf.QPDF_TRUE) - qpdf.qpdf_write(instance) - end - while qpdf.qpdf_more_warnings(instance) ~= 0 do - report("warning: %s",qpdf.qpdf_get_error_full_text(instance,qpdf.qpdf_next_warning(qpdf))) - end - if qpdf.qpdf_has_error(instance) ~= 0 then - report("error: %s",qpdf.qpdf_get_error_full_text(instance,qpdf.qpdf_get_error(qpdf))) - end - qpdf.qpdf_cleanup_p(instance) -end - -- scripts.pdf.info("e:/tmp/oeps.pdf") -- scripts.pdf.metadata("e:/tmp/oeps.pdf") -- scripts.pdf.fonts("e:/tmp/oeps.pdf") @@ -268,11 +360,9 @@ if filename == "" then elseif environment.argument("info") then scripts.pdf.info(filename) elseif environment.argument("metadata") then - scripts.pdf.metadata(filename) + scripts.pdf.metadata(filename,environment.argument("pretty")) elseif environment.argument("fonts") then scripts.pdf.fonts(filename) -elseif environment.argument("linearize") then - scripts.pdf.linearize(filename) elseif environment.argument("exporthelp") then application.export(environment.argument("exporthelp"),filename) else |