= Euclid 1966 =

Euclid (1966), Ji he yuan ben = Jihe yuanben
[http://libcoll.mpiwg-berlin.mpg.de/elib/all_documents/02NT95YF ECHO]

Part of [wiki:WO3], or rather a special work order on its own. 
Sent with DESpecs version 1.1.2. 

Sent: ok/date.
Returned: 2009-01-05, [http://pythia.mpiwg-berlin.mpg.de/department1/mpdl/raw-texts/WO3_Euclid_1966.txt/WO3_Euclid_1966_V1.txt raw text]. 
Came with an unknown characters list (<001> and <002>; see attachment "char code.pdf"). 


== 1.  First Analysis ==

=== Difficulties ===

  * The only text to Formax in Chinese

=== Special Instructions ===

  * see attachment


== 2.  Questions From Formax ==

=== Question ===

Questions concerning figures and tables, see attachments "email.doc" and "berlin.doc".

=== Reply ===

The questions were answered in a few stages, which are not relevant. For the final instructions see "Euclid 1966.txt".

=== Final Instructions ===

Final version of the additional instructions for figures and tables: See attachment.


== 3.  Analysis of the Result ==

We had requested a short sample and were content with the result.

=== Findings ===

Unknown characters: <001> appears once in the text and hasn't been included in "ques.xls". It is a badly printed 轂 (U+8F42). It makes sense that they marked it as an unknown character because only by taking the context into account it becomes clear which character is meant.

<002> appears three times and has already appeared in "ques.xls". In "ques.pdf" we told them to transcribe it as U+2F88D. Although this Unicode code point is not included in most Unicode fonts (it is included in Sun-ExtB), we should make it clear for future work orders that they should find every character in Unicode 5.1.0. (Email from 2009-01-05: "There are Chinese characters that we cannot key, and we use <001> and <002> instead of them. We cannot key the word with code 2f88d, so we use <002> instead of it. We have installed the Unicode font, but we cannot find the CJK compatibility ideographs supplement (u+2F800-U+2FA1F) word, can you help us?")

=== Recommendation ===


== 4.  Post-Processing ==

The first XML version of the text is [http://pythia.mpiwg-berlin.mpg.de/department1/mpdl/raw-texts/WO3_Euclid_1966.txt/WO3_Euclid_1966_V2.txt here].

=== Emendations of the raw text ===

normalize the zero in modern page numbers
{{{
s!○!〇!g;  # white circle U+25CB --> ideographic number zero U+3007
}}}

ignore outdentation in the preface
{{{
if ($line < 153) { s!<p x>!<p>!; } 
}}}

pre-process notes that continue on the next line	
{{{
...
}}}

fill in the unknown characters (MSi)
{{{
s!<001>!轂!g;  # s!<001>!<unknown code="001" unicode="8F42">轂</unknown>!g;
s!<002>!<unknown code="002" unicode="2F88D">庶</unknown>!g; # the actual Unicode character 庶 breaks oXygen
}}}

clarify <?> (the list is not complete!)
{{{
s!<？>!<?>!; # line 811: fullwidth question mark U+FF1F --> ASCII question mark U+003F
s!愈<\?>!愈!g; # MSi: the reading is correct
s!丙、等。<?>而戊丙丁、與甲乙丙、又等。!丙、等。而戊丙丁、與甲乙丙、又等。!; # line 1041
# (line 1041: MSi: It is in the middle of a sentence, but a period at this position is quite common nonetheless.) 
}}}

missing line breaks (the list is not complete!)
{{{
s!小於兩直角。則此二橫直線。!小於兩直角。則此二橫<lb/>直線。!; # line 403; may have to do with the neighboring figure
s!俱小於直角。或幷之小於兩直角。!俱小於直角。或幷之小<lb/>於兩直角。!; # line 404
}}}

normalize the hash in the table
{{{
s!＃!#!g; # fullwidth number sign U+FF03 --> ASCII hash, i.e. number sign U+0023
}}}

move the only table in the text (ECHO p.327) out of its surrounding sentence
{{{
s!却云十六與十二之比例。若!却云十六與十二之比例。!;  # line 4562
s!八與三、及二與四之比例。!若<lb/>八與三、及二與四之比例。!; #line 4573
}}}

misc. emendations
{{{
s!N12<114608657010!N12x114608657010!; # line 5: replace "<" in library stamp junk
s!<pb 六><h>幾何原本　卷一之首</h>!<pb 六><rh>幾何原本　卷一之首</rh>!;  # line 245 (obvious mistake)
s!<h>後支前己正論</h>!<p>後支前己正論</p>!;  # line 2175 (Tian Miao: wrong tag)
if ($line == 2992) { s!<h>第三十四題</h>!<h>第十四題</h>!; }  # line 2992 (obvious mistake)
}}}


=== Further processing steps ===

  * metadata
  * unknown characters
  * figures
  * the table on ECHO p.327
  * ad hoc tagging of book covers, preface, chapters, chapter heads, chapter mains, backmatter
  * page breaks
	  * The ECHO pages 215 to 220 reduplicate the pages 209 to 214 and have been typed only once
  * headings
	  * notes in headings
	  * headings at the lowest level
  * paragraphs
	  * normalize the periods: there should always be a period before </p> and </sm> 
	  (if the period is missing, insert an ASCII period)
	  * turn small text into notes
	  * tag sentences
	  * end each line with <lb/>
	  * outdented paragraphs


=== Some remaining issues ===

  * The metadata may be incorrect in details. 
  * Will the <var> in figures collide with the normal <var>?
  * Replacing <001> by the correct character works fine, but <002> is in a higher plane of Unicode and kills off oXygen, so I have used the simpler standard version of this character.
  * Some <?> have been post-processed already, and I have removed the respective <?> tags because a <?> tag has no value in itself once the line has been checked. I have compiled a list of lines which contain <?> and/or @ and haven't been post-processed yet. ("。</s><s><unsure/>" means that the period is unclear. And one artifact sentence "。</s><s><unsure/>.</s>")
  * The <desc> and <var> in figures have not been used very consistently. (They did not use <cap> at all, but I only know of one figure where it would make sense to use it.)
  * Figures and note have no place attribute.
  * No <num>, <var> (outside of figures), <ptr>, no IDs yet.
  * The parts of problems with more than one part have not been encoded yet.
  * Four books, i.e. four titles, one in the front (attribute n=1) and three in the body (n=1, 2, 3).