使用pdfminer检测pdf的部分

Question

我正在尝试将会议/期刊论文中的 pdf 转换为 .txt 文件。我基本上希望有一个比当前 pdf 更清晰的结构：在句子结束前没有换行符并突出显示论文的各个部分。我目前正在处理的问题是尝试自动检测部分。也就是说，在下图中，我希望能够找到 ABSTRACT、CSS CONCEPT、1 INTRODUCTION、2 THE BODY OF THE PAPER。.

如果目前使用一个简单的想法，它是有效的。我基本上让 pdf miner 完成它的工作，然后使用 NTLK 来查找句子。

def convert_pdf_to_txt(path, year):
    rsrcmgr = PDFResourceManager()
    retstr = StringIO()
    codec = 'utf-8'
    laparams = LAParams()
    device = TextConverter(rsrcmgr, retstr, codec=codec, laparams=laparams)
    fp = open(path, 'rb')
    interpreter = PDFPageInterpreter(rsrcmgr, device)
    password = ""
    maxpages = 0
    caching = True
    pagenos=set()

    for page in PDFPage.get_pages(fp, pagenos, maxpages=maxpages, password=password,caching=caching, check_extractable=True):
        interpreter.process_page(page)

    text = retstr.getvalue()

    sentences = sent_tokenize(text)

    size = len(sentences)
    i = 0
    path = path[:-3]


    output = open("out.txt", 'w')
    for s in sentences:
        s = s.replace("-\n", '') #remove hyphens
        lines = s.split("\n")
        for line in lines:
            if(line.isupper()): #section are only uppercase. 
            #however, other things are also only uppercase hence my errors 
                line = "--SECTION-- " +line
                output.write("\n\n"+line+"\n")
            else:
                output.write(line)
        output.write("\n")
    fp.close()
    device.close()
    retstr.close()

这给了我整个文件的以下输出：

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758SIG Proceedings Paper in LaTeX Format?Extended Abstract†

--SECTION-- G.K.M.

Tobin§Dublin, OhioSean Fogartywebmaster@marysville-ohio.comLars Thørväld¶The Thørväld GroupHekla, Icelandlarst@affiliation.orgCharles PalmerInstitute for Clarity in DocumentationInstitute for Clarity in DocumentationBen Trovato‡Dublin, Ohiotrovato@corporation.comLawrence P. LeipunerBrookhaven Laboratorieslleipuner@researchlabs.orgJohn SmithPalmer Research LaboratoriesSan Antonio, Texascpalmer@prl.comJulius P. KumquatNASA Ames Research CenterMoffett Field, Californiafogartys@amesres.orgThe Thørväld Groupjsmith@affiliation.orgThe Kumquat Consortiumjpkumquat@consortium.netcolumns), a specified set of fonts (Arial or Helvetica and TimesRoman) in certain specified sizes, a specified live area, centeredon the page, specified size of margins, specified column width andgutter size.


--SECTION-- ABSTRACT
This paper provides a sample of a LATEX document which conforms,somewhat loosely, to the formatting guidelines for ACM SIG Proceedings.1Unpublishedworkingdraft.
Notfordistribution.


--SECTION-- CCS CONCEPTS
• Computer systems organization ? Embedded systems; Redundancy; Robotics; • Networks ? Network reliability;

--SECTION-- KEYWORDS
ACM proceedings, LATEX, text taggingACM Reference Format:Ben Trovato, G.K.M.
Tobin, Lars Thørväld, Lawrence P. Leipuner, SeanFogarty, Charles Palmer, John Smith, and Julius P. Kumquat.
1997.


--SECTION-- SIG
Proceedings Paper in LaTeX Format: Extended Abstract.
In Proceedings ofACM Woodstock conference (WOODSTOCK’97).
ACM, New York, NY, USA,5 pages.
https://doi.org/10.475/123_4

--SECTION-- 2 THE BODY OF THE PAPER
Typically, the body of a paper is organized into a hierarchical structure, with numbered or unnumbered headings for sections, subsections, sub-subsections, and even smaller sections.
The command\section that precedes this paragraph is part of such a hierarchy.3LATEX handles the numbering and placement of these headings foryou, when you use the appropriate heading commands aroundthe titles of the headings.
If you want a sub-subsection or smallerpart to be unnumbered in your output, simply append an asteriskto the command name.
Examples of both numbered and unnumbered headings will appear throughout the balance of this sampledocument.
Because the entire article is contained in the document environment, you can indicate the start of a new paragraph with a blankline in your input file; that is why this sentence forms a separateparagraph.


--SECTION-- 1 INTRODUCTION
The proceedings are the records of a conference.2 ACM seeks to givethese conference by-products a uniform, high-quality appearance.
To do this, ACM has some rigid requirements for the format of theproceedings documents: there is a specified format (balanced double?Produces the permission block, and copyright information†The full version of the author’s guide is available as acmart.pdf document‡Dr.
Trovato insisted his name be first.
§The secretary disavows any knowledge of this author’s actions.
¶This author is the one who did all the really hard work.
1This is an abstract footnote2This is a footnotePermission to make digital or hard copies of part or all of this work for personal orUnpublished working draft.
Not for distribution.
classroom use is granted without fee provided that copies are not made or distributedfor profit or commercial advantage and that copies bear this notice and the full citationon the first page.
Copyrights for third-party components of this work must be honored.
For all other uses, contact the owner/author(s).
WOODSTOCK’97, July 1997, El Paso, Texas USA© 2016 Copyright held by the owner/author(s).


--SECTION-- ACM ISBN 123-4567-24-567/08/06.

https://doi.org/10.475/123_4Submission ID: 123-A12-B3.
2018-10-20 12:29.
Page 1 of 1–5.
2.1 Type Changes and Special CharactersWe have already seen several typeface changes in this sample.
You can indicate italicized words or phrases in your text with thecommand \textit; emboldening with the command \textbf andtypewriter-style (for instance, for computer code) with \texttt.
But remember, you do not have to indicate typestyle changes whensuch changes are part of the structural elements of your article;for instance, the heading of this subsection will be in a sans serif4typeface, but that is handled by the document class file.
Take carewith the use of5 the curly braces in typeface changes; they mark thebeginning and end of the text that is to be in the different typeface.
3This is a footnote.
4Another footnote here.
Let’s make this a rather long one to see how it looks.
5Another footnote.
5960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116WOODSTOCK’97, July 1997, El Paso, Texas USAB. Trovato et al.
You can use whatever symbols, accented characters, or nonEnglish characters you need anywhere in your document; you canfind a complete list of what is available in the LATEX User’s Guide[26].
2.2 Math EquationsYou may want to display math equations in three distinct styles:inline, numbered or non-numbered display.
Each of the three arediscussed in the next sections.
Table 1: Frequency of Special CharactersNon-English or Math

--SECTION-- Ø
?$2

--SECTION-- ?
1Frequency Comments1 in 1,0001 in 54 in 5For Swedish namesCommon in mathUsed in business1 in 40,000 Unexplained usagemand, using \cite.
This article shows only the plainest form of the citation comInline (In-text) Equations.
A formula that appears in the2.2.1running text is called an inline or in-text formula.
It is producedby the math environment, which can be invoked with the usual\begin .
.
.
\end construction or with the short form $ .
.
.
$.
You can use any of the symbols and structures, from ? to ?, availablein LATEX [26]; this section will simply show a few examples of intext equations in context.
Notice how this equation: limn?? x = 0,set here in in-line math style, looks slightly different when set indisplay style.
(See next section).
2.2.2 Display Equations.
A numbered display equation—one setoff by vertical space from the text and centered horizontally—isproduced by the equation environment.
An unnumbered displayequation is produced by the displaymath environment.
Again, in either environment, you can use any of the symbolsand structures available in LATEX; this section will just give a coupleof examples of display equations in context.
First, consider theequation, shown as an inline equation above:Some examples.
A paginated journal article [2], an enumeratedjournal article [11], a reference to an entire issue [10], a monograph(whole book) [25], a monograph/whole book in a series (see 2ain spec.
document) [18], a divisible-book such as an anthology orcompilation [13] followed by the same example, however we onlyoutput the series if the volume number is given [14] (so Editor00a’sseries should NOT be present since it has no vol.
no.
), a chapterin a divisible book [37], a chapter in a divisible book in a series[12], a multi-volume work as book [24], an article in a proceedings(of a conference, symposium, workshop for example) (paginatedproceedings article) [4], a proceedings article with all possible elements [36], an example of an enumerated proceedings article [16],an informally published work [17], a doctoral dissertation [9], amaster’s thesis: [5], an online document / world wide web resource[1, 30, 38], a video game (Case 1) [29] and (Case 2) [28] and [27] and(Case 3) a patent [35], work accepted for publication [31], ’YYYYb’test for prolific author [32] and [33].
Other cites might contain’duplicate’ DOI and URLs (some SIAM articles) [23].
Boris / BarbaraBeeton: multi-volume works as books [21] and [20].
Unpublishedworkingdraft.
Notfordistribution.
2.4 TablesBecause tables cannot be split across pages, the best placement forthem is typically the top of the page nearest their initial cite.
To ensure this proper “floating” placement of tables, use the environmenttable to enclose the table’s contents and the table caption.
The contents of the table itself must go in the tabular environment, to bealigned properly in rows and columns, with the desired horizontaland vertical rules.
Again, detailed instructions on tabular materialare found in the LATEX User’s Guide.
Immediately following this sentence is the point at which Table 1is included in the input file; compare the placement of the tablehere with the table in the printed output of this document.
Notice how it is formatted somewhat differently in the displaymath environment.
Now, we’ll enter an unnumbered equation:A couple of citations with DOIs: [22, 23].
Online citations: [38–40].
just to demonstrate LATEX’s able handling of numbering.
and follow it with another numbered equation:x + 1? ? +2(1)(2)limn?? x = 0?i =0?i =0xi =0f2.3 CitationsCitations to articles [6–8, 19], conference proceedings [8] or maybebooks [26, 34] listed in the Bibliography section of your article willoccur throughout the text of your article.
You should use BibTeX toautomatically produce this bibliography; you simply need to insertone of several citation commands with a key of the item cited in theproper location in the .tex file [26].
The key is a short referenceyou invent to uniquely identify each work; in this sample document,the key is the first author’s surname and a word from the title.
Thisidentifying key is included with each item in the .bib file for yourarticle.
The details of the construction of the .bib file are beyond thescope of this sample document, but more information can be foundin the Author’s Guide, and exhaustive details in the LATEX User’sGuide by Lamport [26].
To set a wider table, which takes up the whole width of the page’slive area, use the environment table* to enclose the table’s contentsand the table caption.
As with a single-column table, this widetable will “float” to a location deemed more desirable.
Immediatelyfollowing this sentence is the point at which Table 2 is included inthe input file; again, it is instructive to compare the placement ofthe table here with the table in the printed output of this document.
It is strongly recommended to use the package booktabs [15]and follow its main principles of typography with respect to tables:(1) Never, ever use vertical rules.
(2) Never use double rules.
Submission ID: 123-A12-B3.
2018-10-20 12:29.
Page 2 of 1–5.
117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232SIG Proceedings Paper in LaTeX FormatWOODSTOCK’97, July 1997, El Paso, Texas USATable 2: Some Typical CommandsCommand A Number Comments\author\table\table*100300400AuthorFor tablesFor wider tablesDefinition 2.2.
If z is irrational, then by ez we mean the uniquenumber that has logarithm z:f (x)?(x) = L.(cid:21)log ez = z.
The pre-defined theorem-like constructs are theorem, conjecture, proposition, lemma and corollary.
The pre-defined definition-like constructs are example and definition.
You can add yourown constructs using the amsthm interface [3].
The styles used inthe \theoremstyle command are acmplain and acmdefinition.
Another construct is proof, for example,Proof.
Suppose on the contrary there exists a real number Lsuch thatFigure 1: A sample black and white graphic.
It is also a good idea not to overuse horizontal rules.
Figure 2: A sample black and white graphic that has beenresized with the includegraphics command.
Unpublishedworkingdraft.
Notfordistribution.
lim(cid:20)x???x · f (x)?(x)f (x) = limx?c2.5 FiguresLike tables, figures cannot be split across pages; the best placementfor them is typically the top or the bottom of the page nearest theirinitial cite.
To ensure this proper “floating” placement of figures,use the environment figure to enclose the figure and its caption.
This sample document contains examples of .eps files to bedisplayable with LATEX.
If you work with pdfLATEX, use files in the.pdf format.
Note that most modern TEX systems will convert .epsto .pdf for you on the fly.
More details on each of these are foundin the Author’s Guide.
As was the case with tables, you may want a figure that spans twocolumns.
To do this, and still to ensure proper “floating” placementof tables, use the environment figure* to enclose the figure and itscaption.
And don’t forget to end the environment with figure*, notfigure!
= limx?c?(x)· limx?cf (x)?(x) = 0·L = 0,Thenl = limx?cwhich contradicts our assumption that l (cid:44) 0.


--SECTION-- 3 CONCLUSIONS
This paragraph will end the body of this sample document.
Remember that you might still have Acknowledgments or Appendices;brief samples of these follow.
There is still the Bibliography to dealwith; and we will make a disclaimer about that here: with the exception of the reference to the LATEX book, the citations in this paperare to articles which have nothing to do with the present subjectand are used as examples only.
?

--SECTION-- A HEADINGS IN APPENDICES
The rules about hierarchical headings discussed above for the bodyof the article are different in the appendices.
In the appendix environment, the command section is used to indicate the start ofeach Appendix, with alphabetic order designation (i.e., the first isA, the second B, etc.)
and a title (if you include one).
So, if you needhierarchical structure within an Appendix, start with subsectionas the highest level.
Here is an outline of the body of this documentin Appendix-appropriate form:2.6 Theorem-like ConstructsOther common constructs that may occur in your article are theforms for logical constructs like theorems, axioms, corollaries andproofs.
ACM uses two types of these constructs: theorem-like anddefinition-like.
Here is a theorem:Theorem 2.1.
Let f be continuous on [a, b].
If G is an antiderivative for f on [a, b], then? baHere is a definition:f (t) dt = G(b) ? G(a).
Submission ID: 123-A12-B3.
2018-10-20 12:29.
Page 3 of 1–5.
A.1 IntroductionA.2 The Body of the PaperA.2.1 Type Changes and Special Characters.
A.2.2 Math Equations.
Inline (In-text) Equations.
Display Equations.
A.2.3 Citations.
233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348WOODSTOCK’97, July 1997, El Paso, Texas USAB. Trovato et al.
Figure 4: A sample black and white graphic that has beenresized with the includegraphics command.
A.2.4 Tables.


--SECTION-- A.2.5
Figures.
A.2.6 Theorem-like Constructs.
Figure 3: A sample black and white graphic that needs to span two columns of text.
Unpublishedworkingdraft.
Notfordistribution.
(Nov. 1996).
booktabs.
Mathematical Society.
http://www.ctan.org/pkg/amsthm.
[2] Patricia S. Abril and Robert Plant.
2007.
The patent holder’s dilemma: Buy, sell,or troll?
Commun.
ACM 50, 1 (Jan. 2007), 36–44.
https://doi.org/10.1145/1188913.
1188915[3] American Mathematical Society 2015.
Using the amsthm Package.
American[4] Sten Andler.
1979.
Predicate Path expressions.
In Proceedings of the 6th.


--SECTION-- ACM
SIGACT-SIGPLAN symposium on Principles of Programming Languages (POPL ’79).
ACM Press, New York, NY, 226–236.
https://doi.org/10.1145/567752.567774[5] David A. Anisi.
2003.
Optimal Motion Control of a Ground Vehicle.
Master’s thesis.
[6] Mic Bowman, Saumya K. Debray, and Larry L. Peterson.
1993.
Reasoning AboutNaming Systems.
ACM Trans.
Program.
Lang.
Syst.
15, 5 (November 1993), 795–825. https://doi.org/10.1145/161468.161471[7] Johannes Braams.
1991.
Babel, a Multilingual Style-Option System for Use withRoyal Institute of Technology (KTH), Stockholm, Sweden.
LaTeX’s Standard Document Styles.
TUGboat 12, 2 (June 1991), 291–301.
TeX Users Group, 84–89.
[8] Malcolm Clark.
1991.
Post Congress Tristesse.
In TeX90 Conference Proceedings.
[9] Kenneth L. Clarkson.
1985.
Algorithms for Closest-Point Problems (ComputationalGeometry).
Ph.D. Dissertation.
Stanford University, Palo Alto, CA.
UMI OrderNumber: AAT 8506171.
[10] Jacques Cohen (Ed.).
1996.
Special issue: Digital Libraries.
Commun.


--SECTION-- ACM 39, 11
[11] Sarah Cohen, Werner Nutt, and Yehoshua Sagic.
2007.
Deciding equivalancesamong conjunctive aggregate queries.
J. ACM 54, 2, Article 5 (April 2007),50 pages.
https://doi.org/10.1145/1219092.1219093[12] Bruce P. Douglass, David Harel, and Mark B. Trakhtenbrot.
1998.
Statecarts inuse: structured analysis and object-orientation.
In Lectures on Embedded Systems,Grzegorz Rozenberg and Frits W. Vaandrager (Eds.).
Lecture Notes in ComputerScience, Vol.
1494.
Springer-Verlag, London, 368–394.
https://doi.org/10.1007/3-540-65193-4_29[13] Ian Editor (Ed.).
2007.
The title of book one (1st.
ed.).
The name of the seriesone, Vol.
9.
University of Chicago Press, Chicago.
https://doi.org/10.1007/3-540-09237-4Chicago, Chapter 100. https://doi.org/10.1007/3-540-09237-4[14] Ian Editor (Ed.).
2008.
The title of book two (2nd.
ed.).
University of Chicago Press,[15] Simon Fear.
2005.
Publication quality tables in LATEX.
http://www.ctan.org/pkg/[16] Matthew Van Gundy, Davide Balzarotti, and Giovanni Vigna.
2007.
Catch me, ifyou can: Evading network signatures with web-based polymorphic worms.
InProceedings of the first USENIX workshop on Offensive Technologies (WOOT ’07).
USENIX Association, Berkley, CA, Article 7, 9 pages.
[17] David Harel.
1978.
LOGICS of Programs: AXIOMATICS and DESCRIPTIVE POWER.
MIT Research Lab Technical Report TR-200.
Massachusetts Institute of Technology, Cambridge, MA.
[18] David Harel.
1979.
First-Order Dynamic Logic.
Lecture Notes in Computer Science,Vol.
68.
Springer-Verlag, New York, NY.
https://doi.org/10.1007/3-540-09237-4[19] Maurice Herlihy.
1993.
A Methodology for Implementing Highly ConcurrentData Objects.
ACM Trans.
Program.
Lang.
Syst.
15, 5 (November 1993), 745–770.
https://doi.org/10.1145/161468.161469[20] Lars Hörmander.
1985.
The 

    

Answer 1

要了解为什么会出现误报，您需要更多地了解 PDF。

PDF（与您可能认为的不同）不是 WYSIWYG 格式。事实上，您应该将 PDF 视为指令的容器。它有自己的编程语言，可以告诉查看（或渲染）软件该做什么。

例子：

转到位置 500, 30
将字体设置为 Arial
将字体大小设置为 12
绘制字符 'H' 的字形
转到位置 500, 35
绘制字符 'ello' 的字形

一些指令块在所谓的对象中组合在一起。每个对象都有自己的编号，对象之间可以相互引用。

对象在交叉引用表（称为 XREF 表）中建立索引。

从 PDF 中提取文本时的一些挑战：

• 对象不需要以逻辑阅读顺序出现
• 有时空白是明确写入的，有时是通过简单地移动绘图光标来实现的
• PDF 并不总是嵌入有逻辑结构。因此，虽然您可能能够清楚地看到 2 列布局，但从软件的角度来看，这不太清楚。

我曾在 iText 工作，这是一家专注于 PDF 相关技术的软件公司。并且以您描述的方式从随机文档中以万无一失的方式提取文本是当前软件包根本无法实现的。

有很多异常可能会破坏您的工作流程：

• 出现在文本中的公式
• 桌子
• 具有行跨度和列跨度的表格
• 带有文字环绕的图像
• 脚注
• 标题
• 页脚