Skip to content

Commit

Permalink
Merge pull request #595 from JabRef/acmfetcher
Browse files Browse the repository at this point in the history 
Fixed #545 - ACM fetcher works again
  • Loading branch information
@lenhard
lenhard committed Dec 28, 2015
2 parents 0ac3f6a + 396c96e commit 0b588f5
Show file tree
Hide file tree
Showing 2 changed files with 85 additions and 118 deletions.
1 change: 1 addition & 0 deletions CHANGELOG.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -18,6 +18,7 @@ to [sourceforge feature requests](https://sourceforge.net/p/jabref/features/) by
### Fixed
- Changes in customized entry types are now directly reflected in the table when clicking "Apply" or "OK"
- Reference list generation works for OpenOffice/LibreOffice again, fixes #593
- ACM fetcher works again, fixes #545

### Removed

Expand Down
202 changes: 84 additions & 118 deletions src/main/java/net/sf/jabref/importer/fetcher/ACMPortalFetcher.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -19,8 +19,6 @@
import java.awt.Dimension;
import java.awt.GridLayout;
import java.io.BufferedReader;
import java.io.BufferedWriter;
import java.io.FileWriter;
import java.io.IOException;
import java.io.InputStreamReader;
import java.net.ConnectException;
Expand Down Expand Up @@ -64,41 +62,41 @@ public class ACMPortalFetcher implements PreviewEntryFetcher {
private final UnitFormatter unitFormatter = new UnitFormatter();
private String terms;

private static final String startUrl = "http://portal.acm.org/";
private static final String searchUrlPart = "results.cfm?query=";
private static final String searchUrlPartII = "&dl=";
private static final String endUrl = "&coll=Portal&short=0";//&start=";
private static final String START_URL = "http://portal.acm.org/";
private static final String SEARCH_URL_PART = "results.cfm?query=";
private static final String SEARCH_URL_PART_II = "&dl=";
private static final String END_URL = "&coll=Portal&short=0";//&start=";

private static final String bibtexUrl = "exportformats.cfm?id=";
private static final String bibtexUrlEnd = "&expformat=bibtex";
private static final String abstractUrl = "tab_abstract.cfm?id=";
private static final String BIBTEX_URL = "exportformats.cfm?id=";
private static final String BIBTEX_URL_END = "&expformat=bibtex";
private static final String ABSTRACT_URL = "tab_abstract.cfm?id=";

private final JRadioButton acmButton = new JRadioButton(Localization.lang("The ACM Digital Library"));
private final JRadioButton guideButton = new JRadioButton(Localization.lang("The Guide to Computing Literature"));
private final JCheckBox absCheckBox = new JCheckBox(Localization.lang("Include abstracts"), false);

private static final int perPage = 20; // Fetch only one page. Otherwise, the user will get blocked by ACM. 100 has been the old setting. See Bug 3532752 - https://sourceforge.net/tracker/index.php?func=detail&aid=3532752&group_id=92314&atid=600306
private static final int PER_PAGE = 20; // Fetch only one page. Otherwise, the user will get blocked by ACM. 100 has been the old setting. See Bug 3532752 - https://sourceforge.net/tracker/index.php?func=detail&aid=3532752&group_id=92314&atid=600306
private static final int WAIT_TIME = 200;
private boolean shouldContinue;

// user settings
private boolean fetchAbstract;
private boolean acmOrGuide;

private static final Pattern hitsPattern = Pattern.compile(".*Found <b>(\\d+,*\\d*)</b>.*");
private static final Pattern maxHitsPattern = Pattern.compile(".*Results \\d+ - \\d+ of (\\d+,*\\d*).*");
//private static final Pattern bibPattern = Pattern.compile(".*'(exportformats.cfm\\?id=\\d+&expformat=bibtex)'.*");
private int piv;

private static final Pattern HITS_PATTERN = Pattern.compile("<strong>(\\d+,*\\d*)</strong> results found");
private static final Pattern MAX_HITS_PATTERN = Pattern
.compile("Result \\d+,*\\d* &ndash; \\d+,*\\d* of (\\d+,*\\d*)");

private static final Pattern fullCitationPattern =
Pattern.compile("<A HREF=\"(citation.cfm.*)\" class.*");
private static final Pattern FULL_CITATION_PATTERN = Pattern.compile("<a href=\"(citation.cfm.*)\" target.*");

private static final Pattern idPattern =
Pattern.compile("citation.cfm\\?id=\\d*\\.?(\\d+)&.*");
private static final Pattern ID_PATTERN = Pattern.compile("citation.cfm\\?id=(\\d+)&.*");

// Patterns used to extract information for the preview:
private static final Pattern titlePattern = Pattern.compile("<A HREF=.*?\">([^<]*)</A>");
private static final Pattern monthYearPattern = Pattern.compile("([A-Za-z]+ [0-9]{4})");
private static final Pattern absPattern = Pattern.compile("<div .*?>(.*?)</div>");
private static final Pattern TITLE_PATTERN = Pattern.compile("<a href=.*?\">([^<]*)</a>");
private static final Pattern ABSTRACT_PATTERN = Pattern.compile("<div .*?>(.*?)</div>");
private static final Pattern SOURCE_PATTERN = Pattern.compile("<span style=\"padding-left:10px\">([^<]*)</span>");


@Override
Expand Down Expand Up @@ -126,55 +124,51 @@ public boolean processQueryGetPreview(String query, FetcherPreviewDialog preview
shouldContinue = true;
acmOrGuide = acmButton.isSelected();
fetchAbstract = absCheckBox.isSelected();
int firstEntry = 1;
String address = makeUrl(firstEntry);
String address = makeUrl();
LinkedHashMap<String, JLabel> previews = new LinkedHashMap<>();

try {
URL url = new URL(address);

String page = Util.getResults(url);

int hits = getNumberOfHits(page, "Found", ACMPortalFetcher.hitsPattern);
String resultsFound = "<div id=\"resfound\">";
int hits = getNumberOfHits(page, resultsFound, ACMPortalFetcher.HITS_PATTERN);

int index = page.indexOf("Found");
int index = page.indexOf(resultsFound);
if (index >= 0) {
page = page.substring(index + 5);
index = page.indexOf("Found");
if (index >= 0) {
page = page.substring(index);
}
page = page.substring(index + resultsFound.length());
}

if (hits == 0) {
status.showMessage(Localization.lang("No entries found for the search string '%0'",
terms),
Localization.lang("Search ACM Portal"), JOptionPane.INFORMATION_MESSAGE);
return false;
} else if (hits > 20) {
status.showMessage(
Localization.lang("%0 entries found. To reduce server load, only %1 will be downloaded.",
String.valueOf(hits), String.valueOf(PER_PAGE)),
Localization.lang("Search ACM Portal"), JOptionPane.INFORMATION_MESSAGE);
}

hits = getNumberOfHits(page, "Results", ACMPortalFetcher.maxHitsPattern);

for (int i = 0; i < hits; i++) {
parse(page, 0, firstEntry, previews);
//address = makeUrl(firstEntry);
firstEntry += ACMPortalFetcher.perPage;
}
hits = getNumberOfHits(page, "<div class=\"pagerange\">", ACMPortalFetcher.MAX_HITS_PATTERN);
parse(page, Math.min(hits, PER_PAGE), previews);
for (Map.Entry<String, JLabel> entry : previews.entrySet()) {
preview.addEntry(entry.getKey(), entry.getValue());
}

return true;

} catch (MalformedURLException e) {
e.printStackTrace();
LOGGER.warn("Problem with ACM fetcher URL", e);
} catch (ConnectException e) {
status.showMessage(Localization.lang("Connection to ACM Portal failed"),
Localization.lang("Search ACM Portal"), JOptionPane.ERROR_MESSAGE);
} catch (IOException e) {
status.showMessage(e.getMessage(),
Localization.lang("Search ACM Portal"), JOptionPane.ERROR_MESSAGE);
e.printStackTrace();
LOGGER.warn("Problem with ACM Portal", e);
}
return false;

Expand Down Expand Up @@ -235,68 +229,55 @@ public boolean processQuery(String query, ImportInspector dialog, OutputPrinter
return false;
}

private String makeUrl(int startIndex) {
StringBuilder sb = new StringBuilder(ACMPortalFetcher.startUrl).append(ACMPortalFetcher.searchUrlPart);
sb.append(terms.replaceAll(" ", "%20"));
sb.append("&start=").append(startIndex);
sb.append(ACMPortalFetcher.searchUrlPartII);
private String makeUrl() {
StringBuilder sb = new StringBuilder(ACMPortalFetcher.START_URL).append(ACMPortalFetcher.SEARCH_URL_PART)
.append(terms.replaceAll(" ", "%20")).append(ACMPortalFetcher.SEARCH_URL_PART_II);

if (acmOrGuide) {
sb.append("ACM");
} else {
sb.append("GUIDE");
}
sb.append(ACMPortalFetcher.endUrl);
sb.append(ACMPortalFetcher.END_URL);
return sb.toString();
}


private int piv;


private void parse(String text, int startIndex, int firstEntryNumber, Map<String, JLabel> entries) {
piv = startIndex;
int entryNumber = firstEntryNumber;
while (getNextEntryURL(text, piv, entryNumber, entries)) {
private void parse(String text, int hits, Map<String, JLabel> entries) {
int entryNumber = 1;
while (getNextEntryURL(text, entryNumber, entries) && (entryNumber <= hits)) {
entryNumber++;
}
}

private static String getEntryBibTeXURL(String fullCitation) {
// Get ID
Matcher idMatcher = ACMPortalFetcher.idPattern.matcher(fullCitation);
Matcher idMatcher = ACMPortalFetcher.ID_PATTERN.matcher(fullCitation);
if (idMatcher.find()) {
return idMatcher.group(1);
}
LOGGER.info("Did not find ID in: " + fullCitation);
return null;
}

private boolean getNextEntryURL(String allText, int startIndex, int entryNumber,
private boolean getNextEntryURL(String allText, int entryNumber,
Map<String, JLabel> entries) {
String toFind = "<strong>" + entryNumber + "</strong><br>";
int index = allText.indexOf(toFind, startIndex);
int endIndex = allText.length();
String toFind = "<div class=\"numbering\">";
int index = allText.indexOf(toFind, piv);
int endIndex = allText.indexOf("<br clear=\"all\" />", index);
piv = endIndex;

if (index >= 0) {
piv = index + 1;
String text = allText.substring(index, endIndex);
// Always try RIS import first
Matcher fullCitation =
ACMPortalFetcher.fullCitationPattern.matcher(text);
ACMPortalFetcher.FULL_CITATION_PATTERN.matcher(text);
String item;
if (fullCitation.find()) {
String link = getEntryBibTeXURL(fullCitation.group(1));
String part;
int endOfRecord = text.indexOf("<div class=\"abstract2\">");
if (endOfRecord > 0) {
if (endIndex > 0) {
StringBuilder sb = new StringBuilder();
part = text.substring(0, endOfRecord);

try {
save("part" + entryNumber + ".html", part);
} catch (IOException e) {
e.printStackTrace(); //To change body of catch statement use File | Settings | File Templates.
}

// Find authors:
String authMarker = "<div class=\"authors\">";
Expand All @@ -309,26 +290,31 @@ private boolean getNextEntryURL(String allText, int startIndex, int entryNumber,

}
// Find title:
Matcher titM = ACMPortalFetcher.titlePattern.matcher(part);
Matcher titM = ACMPortalFetcher.TITLE_PATTERN.matcher(text);
if (titM.find()) {
sb.append("<p>").append(titM.group(1)).append("</p>");
}
// Find month and year:
Matcher mY = ACMPortalFetcher.monthYearPattern.matcher(part);
if (mY.find()) {
sb.append("<p>").append(mY.group(1)).append("</p>");

String sourceMarker = "<div class=\"source\">";
int sourceStart = text.indexOf(sourceMarker);
if (sourceStart >= 0) {
int sourceEnd = text.indexOf("</div>", sourceStart + sourceMarker.length());
if (sourceEnd >= 0) {
String sourceText = text.substring(sourceStart, sourceEnd);
// Find source:
Matcher source = ACMPortalFetcher.SOURCE_PATTERN.matcher(sourceText);
if (source.find()) {
sb.append("<p>").append(source.group(1)).append("</p>");
}
}
}

part = sb.toString();
/*.replaceAll("</tr>", "<br>");
part = part.replaceAll("</td>", "");
part = part.replaceAll("<tr valign=\"[A-Za-z]*\">", "");
part = part.replaceAll("<table style=\"padding: 5px; 5px; 5px; 5px;\" border=\"0\">", "");*/
item = sb.toString();
} else {
part = link;
item = link;
}

JLabel preview = new JLabel("<html>" + part + "</html>");
JLabel preview = new JLabel("<html>" + item + "</html>");
preview.setPreferredSize(new Dimension(750, 100));
entries.put(link, preview);
return true;
Expand All @@ -339,9 +325,9 @@ private boolean getNextEntryURL(String allText, int startIndex, int entryNumber,
return false;
}

private static BibEntry downloadEntryBibTeX(String ID, boolean downloadAbstract) {
private static BibEntry downloadEntryBibTeX(String id, boolean downloadAbstract) {
try {
URL url = new URL(ACMPortalFetcher.startUrl + ACMPortalFetcher.bibtexUrl + ID + ACMPortalFetcher.bibtexUrlEnd);
URL url = new URL(ACMPortalFetcher.START_URL + ACMPortalFetcher.BIBTEX_URL + id + ACMPortalFetcher.BIBTEX_URL_END);
URLConnection connection = url.openConnection();

// set user-agent to avoid being blocked as a crawler
Expand All @@ -360,9 +346,9 @@ private static BibEntry downloadEntryBibTeX(String ID, boolean downloadAbstract)

// get abstract
if (downloadAbstract) {
url = new URL(ACMPortalFetcher.startUrl + ACMPortalFetcher.abstractUrl + ID);
url = new URL(ACMPortalFetcher.START_URL + ACMPortalFetcher.ABSTRACT_URL + id);
String page = Util.getResults(url);
Matcher absM = ACMPortalFetcher.absPattern.matcher(page);
Matcher absM = ACMPortalFetcher.ABSTRACT_PATTERN.matcher(page);
if (absM.find()) {
entry.setField("abstract", absM.group(1).trim());
}
Expand All @@ -371,15 +357,12 @@ private static BibEntry downloadEntryBibTeX(String ID, boolean downloadAbstract)

return entry;
} catch (NoSuchElementException e) {
LOGGER.info("Bad Bibtex record read at: " + ACMPortalFetcher.bibtexUrl + ID + ACMPortalFetcher.bibtexUrlEnd,
LOGGER.info("Bad Bibtex record read at: " + ACMPortalFetcher.BIBTEX_URL + id + ACMPortalFetcher.BIBTEX_URL_END,
e);
return null;
} catch (MalformedURLException e) {
LOGGER.info("Malformed URL.", e);
return null;
} catch (ConnectException e) {
LOGGER.info("Cannot connect.", e);
return null;
} catch (IOException e) {
LOGGER.info("Cannot connect.", e);
return null;
Expand All @@ -404,26 +387,19 @@ private String convertHTMLChars(String text) {
*/
private static int getNumberOfHits(String page, String marker, Pattern pattern) throws IOException {
int ind = page.indexOf(marker);
if (ind < 0) {
throw new IOException("Cannot parse number of hits");
}
String substring = page.substring(ind, Math.min(ind + 42, page.length()));
Matcher m = pattern.matcher(substring);
if (!m.find()) {
LOGGER.info("Unmatched! " + substring);
} else {
try {
// get rid of ,
String number = m.group(1);
//NumberFormat nf = NumberFormat.getInstance();
//return nf.parse(number).intValue();
number = number.replaceAll(",", "");
//System.out.println(number);
return Integer.parseInt(number);
} catch (NumberFormatException ex) {
throw new IOException("Cannot parse number of hits");
} catch (IllegalStateException e) {
throw new IOException("Cannot parse number of hits");
if (ind >= 0) {
String substring = page.substring(ind, Math.min(ind + 100, page.length()));
Matcher m = pattern.matcher(substring);
if (m.find()) {
try {
String number = m.group(1);
number = number.replaceAll(",", ""); // Remove , as in 1,234
return Integer.parseInt(number);
} catch (IllegalStateException | NumberFormatException ex) {
throw new IOException("Cannot parse number of hits");
}
} else {
LOGGER.info("Unmatched! " + substring);
}
}
throw new IOException("Cannot parse number of hits");
Expand All @@ -439,22 +415,12 @@ public String getHelpPage() {
return "ACMPortalHelp.html";
}

// This method is called by the dialog when the user has cancelled the import.
public void cancelled() {
shouldContinue = false;
}

// This method is called by the dialog when the user has cancelled or
//signalled a stop. It is expected that any long-running fetch operations
//signaled a stop. It is expected that any long-running fetch operations
//will stop after this method is called.
@Override
public void stopFetching() {
shouldContinue = false;
}

private static void save(String filename, String content) throws IOException {
try(BufferedWriter out = new BufferedWriter(new FileWriter(filename))) {
out.write(content);
}
}
}

0 comments on commit 0b588f5

Please sign in to comment.