[sc-issues] [Issue 48335] drops leading ' inside "" when reading csv
To comment on the following update, log in, then open the issue: http://www.openoffice.org/issues/show_bug.cgi?id=48335 --- Additional comments from [EMAIL PROTECTED] Thu May 5 19:35:42 -0700 2005 --- Good Call. - Please do not reply to this automatically generated notification from Issue Tracker. Please log onto the website and enter your comments. http://qa.openoffice.org/issue_handling/project_issues.html#notification - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
[sc-issues] [Issue 48335] drops leading ' inside "" when reading csv
To comment on the following update, log in, then open the issue: http://www.openoffice.org/issues/show_bug.cgi?id=48335 --- Additional comments from [EMAIL PROTECTED] Mon May 2 19:26:20 -0700 2005 --- Firstly, thanks I now know where we are, I think I may even understand what you think 'one issue one issue report' means and will endeavour to comply. (probably by making feature requests.) Alan === first revisit how many issues this is for a moment. (pls ignore/skip to ===) It is your world, from now on, after this one, I will not add amy more comments to this issue re: anything other than ' inside "" unless asked to. In my view it is not reasonable to consider this issue piecemeal especially when I think one line of code could fix the whole issue. Hence I thought this is a bug fix rather than a feature request. I think the issue may however have moved focus, you say "Up to now it is a clear strategy to handle CSV content identical to input via the keyboard." In that case the bug is that when saving the csv file the data is not correctly reverse translated from text cell 'Here is my text' into "''Here is my text'" If in reading *and* writing csv (as you said) "CSV content identical to input via the keyboard" then surely by symmetry fields must be translated on writing into the text that would have to be inputed to produce this field. While that is a self consistent plan I would refrain from calling it csv. Perhaps that format is CSUI comma separated user input. === Now to the topic at hand. (repeats the above a bit I am afraid, but I did ask you to skip that bit as an aside.) I accept that it is possible to, decide that treating input from csv identically to user typed input is one strategy, and although it is not what I want today it is a strategy. In that case however I would expect that the process that output to csv mirrored this concept as far as is possible. Thus when a text cell, is output to a file, it would be printed out as the text that should be input to achieve the same result. Thus, a text cell would have a single ' prepended on output. (Dont do this it is a bad idea, which is why it is the reading code I claimed was faulty.) Consider the following user story. A User types ''Here is my text' into open office in cell a1. (first ' removed) cell a1 is text and contains 'Here is my text'. (all good so far) It is saved to file. (Sure on closing the document we get a warning that some information may have been lost. I expect formulas and all sorts of stuff to go, but I do expect it (OOo) to perform as well as the evil alternatives.) User Opens the csv file in vi shows "'here is my text'" as contents. Good thinks the User, there is my data. User Open the csv file in Open Office and the leading ' is missing. Hmm. Yes Open Office can read csv files, but it cant read the ones it wrote. In particular it deletes, ' inside "" reads in "1.0" as a numeric 1 Yes it happens with dates too type '23 may 2005 into a1 save as csv Look in vi Yep there it is as "23 may 2005". Load the csv file with, OOo and save as csv again (because you want to change some other field) and it has changed a1 to 23/05/2003. Yes if the whole column is text we can specify that in the GUI, each and every time we open the file. If OOo had the same file defintion for reading as writing csv then it would respect the "" surrounding any field and treat it as a text cell. It would communicate this intention to the GUI by prepending a ' (patch 1)(or through a modified interface patch 2) If OOo had the same file defintion for writing as reading csv (note order change) then when writing csv it would write down what a user should type in ,in order to get this cell contents. ie it would write down a leading ' inside the "". (note this is not an advocated feature and has real problems) As it appears this is not a bug, but perhaps a future requirement, I would suggest you (OO org) make it a priority. I think it is a serious issue when OOo is less compatable with the open csv standard than the evil propriatory alternative. (yes there is no official standard, but there is standard practise and there is the follwoign measure of merit.) OOo is less compatable because there are many fewer files that it can export to csv and then read back in and get the same content. If the evil alternative did this I would shout abuse. If and when this requirement is to be spec'd and made a feature I am prepared to put my neck on the line and contribute, and probably answer any complaints about what I think csv format is, or whether my code works as intended. (there is a rasonable chance I will have a local working version by then anyway. ) Permanent Contact Details. alan aiaware.com will (with the obvious change,insert @) when the server is fixed reach me. alan aiaware_nimda.hotpop.com is my domain contact and is getting spammed and will be moved as soon as I fix my email server. To
[sc-issues] [Issue 48335] drops leading ' inside "" when reading csv
To comment on the following update, log in, then open the issue:
http://www.openoffice.org/issues/show_bug.cgi?id=48335
--- Additional comments from [EMAIL PROTECTED] Fri Apr 29 01:40:12 -0700
2005 ---
Note this is not part of my problem, my problem is
drops leading ' inside "" when reading csv
I have always found that code faults cluster and it is easiest to fix
everything that is wrong with the localised piece of functionality all at once.
Note OOo also interprets csv fields with leading '=' as special
(the one liner also fixes that)
eg ...,"=a1",... is read as a formula
(That might make sense but formulas are not exported to csv that way.)
This is actually a potentially good feature.
IF
...,=a1,... is interpreted as a formula AND
...,"=a1",... is interpretd as text
AND the GUI for export as CSV is given a check box to 'export formulae'
THEN
export csv with export formulae checked would write
...,=a1,... to the file
export csv with export formulae NOT checked would write
...,"Contents of cell a1",... to the file (as it does now)
I have in my time written rather a lot of robust CSV and related parsing IO code
I could, if there was need, write a serious spec/defintion of CSV IO
and provide a test set of data to thoroughly exercise the algorithm.
Code that decides ' && = are special
BOOL ScColumn::SetString( SCROW nRow, SCTAB nTab, const String& rString )
{ ...
if ( !bIsLoading ) { ...X... }
...
if ( cFirstChar == '=' )
{
...
} else if ( cFirstChar == '\'')
...
}
I suspect we/you cant mess easily with the code at ...X... because my bet is
some of the other format importers probably depend on it. Looks like the code
needs a good refactoring ... its already been patched.
-
Please do not reply to this automatically generated notification from
Issue Tracker. Please log onto the website and enter your comments.
http://qa.openoffice.org/issue_handling/project_issues.html#notification
-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
[sw-issues] [Issue 44511] random style changes after inserting paragraph
To comment on the following update, log in, then open the issue: http://www.openoffice.org/issues/show_bug.cgi?id=44511 --- Additional comments from [EMAIL PROTECTED] Fri Apr 29 00:16:08 -0700 2005 --- I am only wondering if what I know is relevant, but as I have the information and you dont I thought I would throw what I know in the pot too. You mention the use of converting to word format. (did you use word?) I have used word at work and had the same symptoms, but never used the document with OOo. Makes me wonder if the 'bug' is something word does at wrong time 1 but remains invisible until at time 2 an application (Word or OOo) correctly inserts a \n and the symptom shows up. When it happens to a word document, in word, I can make the problem go away by copying a good paragrah mark from somewhere else and pasting it over all the paragaph marks near where the problem is. The possibility of changing the docment state to suppress the word bug makes me think word can get a document to a faulty but visually correct state. ie I guess I am recommending a close inspection of the document model in the debugger with this doc loaded into memory. The behaviour I have seen in word makes me think there is a paragraph mark that is missing its paragraph type specification. - Please do not reply to this automatically generated notification from Issue Tracker. Please log onto the website and enter your comments. http://qa.openoffice.org/issue_handling/project_issues.html#notification - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
[sc-issues] [Issue 48335] drops leading ' inside "" when reading csv
To comment on the following update, log in, then open the issue:
http://www.openoffice.org/issues/show_bug.cgi?id=48335
--- Additional comments from [EMAIL PROTECTED] Thu Apr 28 09:33:05 -0700
2005 ---
I have not yet compiled the code but inspection finds this code
which from my reading shows the behaviours depicted by the examples.
The line aCell.Append("'",1); //< Insert this
will probably fix the symtoms aka patch 1
Changing the interface to pDoc to accept one more bool that is set true
when this function parses a quoted "" text field would I think be a better
solution but require more lines of code to change.
pDoc->SetString( nCol, nRow, aRange.aStart.Tab(), aCell, bIsString );
aka patch 2.
Code from
impex.cxx
BOOL ScImportExport::Text2Doc( SvStream& rStrm )
... deleted ...
while( *p )
{
aCell.Erase();
if( *p == cStr )
{
aCell.Append("'",1); //< Insert this
p = lcl_ScanString( p, aCell, cStr,
DQM_KEEP );
while( *p && *p != cSep )
p++;
if( *p )
p++;
}
else
{
const sal_Unicode* q = p;
while( *p && *p != cSep )
p++;
aCell.Assign( q, p - q );
if( *p )
p++;
}
if (ValidCol(nCol) && ValidRow(nRow) )
{
if( bSingle )
{
if (nCol>nEndCol) nEndCol =
nCol;
if (nRow>nEndRow) nEndRow =
nRow;
}
if( bData && nCol <= nEndCol && nRow <=
nEndRow )
pDoc->SetString( nCol, nRow,
aRange.aStart.Tab(), aCell );
}
...
-
Please do not reply to this automatically generated notification from
Issue Tracker. Please log onto the website and enter your comments.
http://qa.openoffice.org/issue_handling/project_issues.html#notification
-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
[sc-issues] [Issue 48335] drops leading ' inside "" when reading csv
To comment on the following update, log in, then open the issue: http://www.openoffice.org/issues/show_bug.cgi?id=48335 User alanchristiansen changed the following: What|Old value |New value Status|CLOSED|UNCONFIRMED Resolution|INVALID | --- Additional comments from [EMAIL PROTECTED] Thu Apr 28 07:37:21 -0700 2005 --- I believe that this report is valid and addresses one problem. The problem is drops leading ' inside "" when reading csv The included file demostrates this property. The ... in the body of the text demostrates this property. In attempting to produce definitive examples by black box testing I found several other characteristics of the software, that I really do believe are the same problem, where 'problem' is defined as 1 fault in the code needing remedy. Assuming std model view controller terminology. There will be two modules inside the code, one responsible for parsing CSV and one for the 'model' of the spreadsheet. The interface from the control (GUI) to the model uses one method to specify that a cell is text. It probably does this implicitly by just passing the string and if the string starts with ' it means it is a text cell even if the contents look numeric. I believe you will find that the CSV reading code incorrectly uses that interface and hence fails to correctly comunicate its intentions to the model. I presented several alternative potential methods to fix this, the first made the CSV conform to the interface made for the GUI the second and I think I indicated the preferred (in my view) method which modifies the model interface to allow code to specify to the model explicitly if this cells contents are text. - Please do not reply to this automatically generated notification from Issue Tracker. Please log onto the website and enter your comments. http://qa.openoffice.org/issue_handling/project_issues.html#notification - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
[www-issues] [Issue 48379] New - bad PR for Open Office, with new clients.
To comment on the following update, log in, then open the issue: http://www.openoffice.org/issues/show_bug.cgi?id=48379 Issue #:|48379 Summary:|bad PR for Open Office, with new clients. Component:|www Version:|OOo 1.1 Platform:|All URL:| OS/Version:|All Status:|UNCONFIRMED Status whiteboard:| Keywords:| Resolution:| Issue type:|DEFECT Priority:|P3 Subcomponent:|openoffice.org IssueTracker Assigned to:|[EMAIL PROTECTED] Reported by:|alanchristiansen --- Additional comments from [EMAIL PROTECTED] Thu Apr 28 07:04:52 -0700 2005 --- Recommended Action: mark this issue as invalid too, close it, and have a nice day. :) === 1 problem at an abstract level The most abstract description of this issue is my concern about the image of Open Office as presented to people like me who are widely experienced in IT, and try to just flip into Open Office and be productive, report whatever bugs they find like good citizens. Being mad as hell at their former software supplier they could even feel useful contributing to the solution instead of paying for the problem. The best example of what I hoped to find here was what happened when Installed Open BSD and my 12 yr old son browsed some of the doco, "Hey" he said in suprised tone, "Thats clear, precise, *and* it sound slike it is trying to help." He was used to the mudddy confused buck passing help he got from our previous software supplier. So far I cant say the same about here. A question is why do I care about Open Offices image? I care cos I know which side my bread should be buttered on and I very much want that side to come out on top. === 1 problem The same one at a concrete example based level Problem: Issue 48335 is now closed and marked invalid, I dont think it is invalid, I think it addresses 1 issue, defined by its summary "drops leading ' inside "" when reading csv" I doubt I could have been more precise or succinct in this summary. Problem: Oh no two problems, this report is invalid, kill it now before it spreads. (sorry I just had to say that) Problem/Suggestion: I did something creative I read some random example issues, I see that my style does not match expectations, it 48335 was however one issue. I may well have missed it but it would probably be a good thing to try and get newly registered users to browse some random issue reports, so that the newbies to your group can absorb local customs and not scare the natives. Problem/Suggestion: When obviously new people join your social circle and submit reports that you believe address multiple issues, perhaps you as a group should adress their first issue, and reply that they need to submit the rest separately. AT the very least this provides one concrete example of what you think an issue is. (and as in the BSD doco (described above), you would be trying to help.) It is true that inside 48335 it provides copious details and examples that I believe should to a programmer familar with the code identify exactly where to look, I believe I also even suggested a good viable solution. (and lesser alternatives) My problem is that considering the email reply I got I see no way forward that gets me version of the code, in any release that addresses the issue I raised. The minimum effort path that I see for me to go forward is to find access to a stable version of the source, institute a local branch on my machine, and apply my patches to each version of OOffice that I choose to use. Works for me My problem is that I thought we were supposed to be cooperating together, and making many hands make light work. It is possible my 48335 issue report was unclear, I have been known to express myself poorly in English as opposed to my native tongue (code). If I with the experience I have cannot submit a valid issue, then I expect none of the people I thought to recommend swapping to Open Office will be able to either. I suspect that I would thus lose their respect for me if I lead them there. Now while the evil alternative is no better it is well greased and its not on my head if I let the locals here go on using it. I hoped to be,(understood I was) in a position to recomend to my less literate computer user friends that they could finally abandon the evil empire and that open source software really answered their needs now that it had breached the incompatibility wall. Sad, but well fed, thanks for all the fish (open src code) Alan - Please do not reply to this automatically generated notification from Issue Tracker. Please log o
[sc-issues] [Issue 48335] drops leading ' inside "" when reading csv
To comment on the following update, log in, then open the issue: http://www.openoffice.org/issues/show_bug.cgi?id=48335 --- Additional comments from [EMAIL PROTECTED] Thu Apr 28 00:14:21 -0700 2005 --- Hmmm. I was very sure that I got OO on reading ...,"Foo" "Bar",... to imagine a comma between them, but the test file I have sent you instead totally ignores "Bar" but as I said I consider this to be an invalid CSV file, and so the only issue might be that perhaps Open Office should complain rather than silently ignore the "Bar" in the input file. YMMV Alan - Please do not reply to this automatically generated notification from Issue Tracker. Please log onto the website and enter your comments. http://qa.openoffice.org/issue_handling/project_issues.html#notification - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
[sc-issues] [Issue 48335] drops leading ' inside "" when reading csv
To comment on the following update, log in, then open the issue: http://www.openoffice.org/issues/show_bug.cgi?id=48335 User alanchristiansen changed the following: What|Old value |New value Attachment data| |Created an attachment | |(id=25589) Example file | |that loads, badly, pls see | |additional comments --- Additional comments from [EMAIL PROTECTED] Thu Apr 28 00:10:24 -0700 2005 --- Created an attachment (id=25589) Example file that loads, badly, pls see additional comments - Please do not reply to this automatically generated notification from Issue Tracker. Please log onto the website and enter your comments. http://qa.openoffice.org/issue_handling/project_issues.html#notification - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
[sc-issues] [Issue 48335] New - drops leading ' inside "" when reading csv
To comment on the following update, log in, then open the issue: http://www.openoffice.org/issues/show_bug.cgi?id=48335 Issue #:|48335 Summary:|drops leading ' inside "" when reading csv Component:|Spreadsheet Version:|OOo 1.1.4 Platform:|PC URL:| OS/Version:|All Status:|UNCONFIRMED Status whiteboard:| Keywords:| Resolution:| Issue type:|DEFECT Priority:|P3 Subcomponent:|open-import Assigned to:|spreadsheet Reported by:|alanchristiansen --- Additional comments from [EMAIL PROTECTED] Wed Apr 27 23:58:36 -0700 2005 --- For this data Defintion: The data ...,"Date","'""'dd MMM '""'","'""'dd MMM '""'",... when loaded into spread sheet as csv (comma separators) " (text delimeter) drops the leading ' NOTE: It does preview correctly, it does not wind up in the spreadsheet correctly. Worst case problem: Open file (comma sep " text delim) ...,"Date","'""'dd MMM '""'","'""'dd MMM '""'",... Change nothing, save it, Open it again and its different. Workaround none. Cant read csv data with leading ' on the fields. Probably Related problem : Open file (comma sep " text delim) "1.0",1.0,"1.0" , "WS B4Q"," Trailing WS on line" Change nothing, save it, Open it again and its different. 1,1,1," ""WS B4Q"""," Trailing WS on line" Workaround none. Cant read csv data with "" on the fields to indicate numeric looking stuff is text. Note I suspect that , " should as it does be defined as meaning a string starting with a space containing a quote. Related problem by way of contrast: Open file (comma sep " text delim) ...,"Foo" "Bar",... Change nothing, save it, Open it with a text editor and it is different. Ok so the file is syntactically invalid, Open office *silently* imagines the comma between ...,"Foo","Bar",... and when the file is saved wites it out. Workaround garbage in garbage out, dont feed it (OO) garbage.. Being a programmer (who has not seen any src yet...) I offer some guesses. Use the GUI. Into a cell Type 'text or Type '1.0 in both circumstances the GUI understands (the meta info) that the user meant the cell to be text. Use the GUI. Into a cell type 1.0 in this circumstance the GUI understands (the meta info) that the user meant the cell to be a number. When reading a CSV file, there is (currently?) no meta info channel. (that is implemented correctly?) Patch 1: (code smell, of bandages) The code reading the csv file should for every field read from the file that has a leading ' add a second ' before inserting it. *AND* For every field read from the file that is wrapped in "" prepend ' to indicate that the meta channel said it was text. (Note if ' is added to every field then all numeric fields will turn into text cells.) Patch 2: The existing function used for inputting cell char data supplied by a human should not be used for inputting cell char data supplied by a file a new related function that does not strip leading ' should be added to the interface. This new related function has a second boolean parameter to indicate if the field should be entered as a text cell. The file code says true if it was wrapped in "" The existing function says true if it stripped the leading ' typed by a human. Patch 3: The devils advocates alternative The CSV writing code needs to be symetrical with the reading code. All text cells written to csv files need to be both wrapped in "" and prepended with '. Patch 4: The devils advocates alternative The CSV writing code needs to be symetrical with the reading code. All text cells starting with ' or that llok numeric and written to csv files need to be prepended with a '. As open office is not mickey mouse, open office should not redfine well known standards such as csv on either writing (patch 3&4) or reading (current bug) note on contacting me, my email is a bit unwell alan aiaware.com will (with the obvious change,insert @) when it is fixed reach me. alan aiaware_nimda.hotpop.com is getting spammed and will be moved as soon as I fix my email. - Please do not reply to this automatically generated notification from Issue Tracker. Please log onto the website and enter your comments. http://qa.openoffice.org/issue_handling/project_issues.html#notification - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
