README (8627B)
1 These files are binary test files that were publicly available 2 on the Internet and were reproduced here to test the logic of 3 the respective plugin. They are not under the "GPL", but 4 assumed to be reproducable (especially in this context) as they 5 are used for compatibility testing and their content is often 6 so trivial that they may not be copyrightable in the first place. 7 In other cases, they were distributed liberally by the original 8 authors. 9 10 11 Provenance of the MS Office samples 12 =================================== 13 14 The files below were added for the `msoffice' plugin and for the 15 extensions to the `ole2' plugin that extract the names of the people 16 who edited a document (https://bugs.gnunet.org/view.php?id=2096). 17 Because that is precisely the kind of information the plugins are 18 meant to surface, the provenance and the licensing status of each 19 file is recorded here explicitly. 20 21 Files taken from Govdocs1 22 ------------------------- 23 24 Govdocs1 is a research corpus of about one million documents that were 25 collected by Simson Garfinkel et al. from web servers in the .gov 26 domain, and which is published for research use at 27 https://digitalcorpora.org/corpora/file-corpora/files/ with the 28 statement that the files "are freely available for research and may be 29 (to the best of our knowledge) freely redistributed". Works of the 30 U.S. federal government are not subject to copyright protection in the 31 United States (17 U.S.C. section 105), so these files are believed to 32 be in the public domain there; as with the pre-existing test files 33 above, they are reproduced here for compatibility testing. 34 35 Each file was renamed; the original number is the file's identifier 36 within the corpus and is given below. 37 38 msoffice_word.docx govdocs1 732704.docx 39 Word 2007 document, "Database Deployment 40 Request". Carries core and extended 41 properties and one tracked insertion whose 42 timestamp is in local time, five hours off 43 the UTC modification date. 44 45 msoffice_excel.xlsx govdocs1 610528.xlsx 46 Excel 2007 workbook (National Park 47 Service). 48 49 msoffice_powerpoint.pptx govdocs1 392790.pptx 50 PowerPoint 2007 presentation (Department 51 of Veterans Affairs). 52 53 msoffice_biff4.xls govdocs1 149599.xls 54 An Excel worksheet written as a bare BIFF 55 record stream, i.e. without an OLE2 56 container, which is how Excel 2 to Excel 4 57 stored files. Its only meta data is the 58 user name "FCC" in the WRITEACCESS record 59 of the File Protection Block. 60 61 ole2_powerpoint.ppt govdocs1 167115.ppt 62 PowerPoint 97-2003 presentation (FDA, 63 Center for Drug Evaluation and Research). 64 Its `Current User' stream names 65 "FDA.CDER", which differs from both the 66 author and the last-saved-by recorded in 67 the summary information -- a good example 68 of why the stream is worth reading. 69 70 All five name organisations or role accounts rather than private 71 individuals. That was a deliberate selection criterion: many 72 documents in the corpus name identifiable people, and there is no 73 reason to republish those here when a functionally equivalent sample 74 does not. 75 76 Synthetic files 77 --------------- 78 79 The public corpora predate Office 2013, so they contain none of the 80 parts that tie an author name to a directory identity, and no 81 documents that had been circulated for review. The following two 82 files were therefore written by hand for this test suite. They are 83 valid OPC packages, they are part of libextractor and covered by its 84 licence, and every name and address in them is fictitious 85 (example.com is reserved for documentation by RFC 2606). 86 87 msoffice_review.docx Word document with the `_AuthorEmail' and 88 `_AuthorEmailDisplayName' custom 89 properties that Outlook leaves behind when 90 a document is sent out for review, a 91 comment, tracked changes by two authors, 92 and a word/people.xml part mapping author 93 names to an e-mail address and to a 94 Windows SID. 95 96 msoffice_shared.xlsx Workbook with <fileSharing userName="..."> 97 (the OOXML descendant of the BIFF 98 FILESHARING record), a classic cell 99 comment author list, the threaded comment 100 `persons' part, and a shared workbook 101 revision log naming who changed the 102 workbook and when. 103 104 105 Provenance of the RTF samples 106 ============================= 107 108 The files below were added for the `rtf' plugin, which extracts the 109 same class of information from Rich Text Format documents: who wrote 110 them, who last saved them, who made each tracked change and when. 111 They come from the same Govdocs1 corpus described above, under the 112 same terms, and were selected under the same criterion -- they name 113 organisations, role accounts or login names rather than identifiable 114 private individuals. 115 116 rtf_nih.rtf govdocs1 003332.rtf 117 National Institute of Neurological 118 Disorders and Stroke. `\author' and 119 `\operator' are both the role account 120 "NINDS", `\*\company' is "NIH". 121 122 rtf_fda.rtf govdocs1 942656.rtf 123 Announcement of a meeting of the Medical 124 Imaging Drugs Advisory Committee. Written 125 by the same "CDER USER" account as the 126 PowerPoint sample above. 127 128 rtf_nasa.rtf govdocs1 806272.rtf 129 Goddard Space Flight Center. Created 130 under the login "WLBEHNKE" but last saved 131 by "INTELLISOURCE", the contractor running 132 the office -- which is why `\author' and 133 `\operator' are reported separately. 134 135 rtf_fta.rtf govdocs1 882751.rtf 136 Federal Transit Administration document 137 produced by scanning software, which left 138 its own name in `\author' and a custom 139 document property in `\*\userprops'. 140 141 rtf_doe_mac.rtf govdocs1 534437.rtf 142 Review of the Yucca Mountain Igneous 143 Consequences Peer Review Panel report. 144 Written with Word for the Macintosh, which 145 says so by declaring the Mac OS Roman 146 character set; the apostrophe in the title 147 is encoded as `\'d5', which decodes 148 correctly only if that declaration is 149 honoured. 150 151 Synthetic file 152 -------------- 153 154 rtf_review.rtf Written by hand for this test suite, and 155 hence part of libextractor and covered by 156 its licence. Every name and address in it 157 is fictitious (example.org is reserved for 158 documentation by RFC 2606). It covers the 159 paths that carry personal data and that 160 therefore should not be exercised with a 161 real document: a `\*\revtbl' revision 162 table with two authors, the `\revauth', 163 `\revauthdel' and `\crauth' flavours of 164 revision mark with their packed time 165 stamps, `\*\atnid' / `\*\atnauthor' / 166 `\*\atndate' for a comment, the Outlook 167 `_AuthorEmail' review properties, a 168 `\*\template' path from the authoring 169 machine, and both the `\uN' and the 170 `\'hh' character escapes.