From: Kevin M. Rosenberg Date: Tue, 9 Apr 2013 19:36:23 +0000 (-0600) Subject: Add entrez package, not currently working X-Git-Url: http://git.kpe.io/?p=pubmed.git;a=commitdiff_plain;h=5ce187bd43a9280f16a315121967f39b9d2d1e10;ds=sidebyside Add entrez package, not currently working --- diff --git a/README b/README index 7334158..18755e9 100644 --- a/README +++ b/README @@ -1,9 +1,9 @@ -This is the Common Lisp PubMed package. +This is the Common Lisp Entrez package. This package is written and Copyright (C) 2000-2002 by Kevin M. Rosenberg -The web site for this package is http://pubmed.b9.com/ +The web site for this package is http://entrez.b9.com/ Prerequisites @@ -19,17 +19,17 @@ Prerequisites Installation ============ - Debian: - evaluate (require 'pubmed) + evaluate (require 'entrez) - Other Operating Systems: - 1. Install the ASDF, KMRCL, and PubMed packages into directories + 1. Install the ASDF, KMRCL, and Entrez packages into directories of your choice. 2. Load asdf.lisp. - 3. Push the names of the directories containing the KMRCL and PubMed + 3. Push the names of the directories containing the KMRCL and Entrez files on to the asdf:*central-registry* list. 2. Load kmrcl.asd from the KMRCL package. - 3. Load pubmed.asd from the PubMed package. - 4. Evaluate (asdf:oos 'asdf:load-op 'pubmed). + 3. Load entrez.asd from the Entrez package. + 4. Evaluate (asdf:oos 'asdf:load-op 'entrez). USAGE diff --git a/debian/README.Debian b/debian/README.Debian index 841ad31..bf068dd 100644 --- a/debian/README.Debian +++ b/debian/README.Debian @@ -1,6 +1,6 @@ To use this package in Debian, execute -(clc:require 'pubmed) +(clc:require 'entrez) in your Lisp environment. diff --git a/debian/changelog b/debian/changelog index 0daa711..81b0a0d 100644 --- a/debian/changelog +++ b/debian/changelog @@ -1,102 +1,5 @@ -cl-pubmed (2.1.3-5) unstable; urgency=low - - * Build with debhelper extension dh-lisp - * control: Add Vcs-Browser field. - - -- Kevin M. Rosenberg Mon, 03 Aug 2009 18:42:04 -0600 - -cl-pubmed (2.1.3-4) unstable; urgency=low - - * control: Really add Vcs-Git and Homepage fields. - - -- Kevin M. Rosenberg Sun, 02 Aug 2009 01:44:07 -0600 - -cl-pubmed (2.1.3-3) unstable; urgency=low - - * debian/watch: New file - * debian/control: Require debhelper 7. Change to new lisp section. - Add Vcs-Git and Homepage fields. - * debian/compat: Update to version 7 - * debian/rules: Update for debhelper 7, architecture-independent build - * debian/{prerm,postrm}: Remove path from binary function - - -- Kevin M. Rosenberg Sun, 02 Aug 2009 01:41:26 -0600 - -cl-pubmed (2.1.3-2) unstable; urgency=low - - * debian/rules: don't export DH_COMPAT - * debian/control: Use build-depends, update standards-version - - -- Kevin M. Rosenberg Sun, 09 Sep 2007 19:28:07 -0600 - -cl-pubmed (2.1.3-1) unstable; urgency=low - - * New upstream - - -- Kevin M. Rosenberg Sat, 17 Apr 2004 12:24:07 -0600 - -cl-pubmed (2.1.2-1) unstable; urgency=low - - * Support proxy server - - -- Kevin M. Rosenberg Thu, 15 Apr 2004 00:47:01 -0600 - -cl-pubmed (2.1.1-1) unstable; urgency=low - - * New upstream - - -- Kevin M. Rosenberg Thu, 12 Jun 2003 06:16:58 -0600 - -cl-pubmed (2.1.0-1) unstable; urgency=low +cl-entrez (0.1-1) unstable; urgency=low * New upstream - -- Kevin M. Rosenberg Fri, 6 Jun 2003 16:04:55 -0600 - -cl-pubmed (2.0.2-1) unstable; urgency=low - - * Improve .asd file - - -- Kevin M. Rosenberg Tue, 6 May 2003 10:24:12 -0600 - -cl-pubmed (2.0.1-1) unstable; urgency=low - - * Remove 'load-compiled-op from .asd file - - -- Kevin M. Rosenberg Mon, 2 Dec 2002 11:42:38 -0700 - -cl-pubmed (2.0-1) unstable; urgency=low - - * Rename classes and accessors to be more friendly for importing into - another package. Export more functions. - * Add article-equal function - * Add server-error condition - * Fix bug - - -- Kevin M. Rosenberg Thu, 31 Oct 2002 07:07:34 -0700 - -cl-pubmed (1.2-1) unstable; urgency=low - - * Fix extraneous colon bug and change in XML tag from - Descriptor to DescriptorName - - -- Kevin M. Rosenberg Wed, 30 Oct 2002 19:01:36 -0700 - -cl-pubmed (1.1.1-1) unstable; urgency=low - - * README: Add prerequisites and installation instructions. - - -- Kevin M. Rosenberg Wed, 30 Oct 2002 18:34:56 -0700 - -cl-pubmed (1.1-1) unstable; urgency=low - - * Fix bug in .asd file - - -- Kevin M. Rosenberg Wed, 30 Oct 2002 14:18:11 -0700 - -cl-pubmed (1.0-1) unstable; urgency=low - - * Initial release (closes: 166482) - - -- Kevin M. Rosenberg Tue, 22 Oct 2002 09:57:38 -0600 - + -- Kevin M. Rosenberg Tue, 09 Apr 2013 13:34:58 -0600 diff --git a/debian/control b/debian/control index fca7f11..e4de648 100644 --- a/debian/control +++ b/debian/control @@ -1,18 +1,18 @@ -Source: cl-pubmed +Source: cl-entrez Section: lisp Priority: optional Maintainer: Kevin M. Rosenberg Build-Depends-Indep: dh-lisp Build-Depends: debhelper (>= 7.0.0) Standards-Version: 3.8.2.0 -Homepage: http://files.b9.com/pubmed/ -Vcs-Git: git://git.b9.com/pubmed.git -Vcs-Browser: http://git.b9.com/?p=pubmed.git +Homepage: http://files.b9.com/entrez/ +Vcs-Git: git://git.b9.com/entrez.git +Vcs-Browser: http://git.b9.com/?p=entrez.git -Package: cl-pubmed +Package: cl-entrez Architecture: all Depends: ${misc:Depends}, cl-kmrcl, cl-aserve -Description: Common Lisp package to query Pubmed medical literature database - This library has functions for querying the PubMed medical literature +Description: Common Lisp package to query Entrez medical literature database + This library has functions for querying the Entrez medical literature database and parsing the XML results into Common Lisp objects. This package uses the cl-aserve for HTTP fetches. diff --git a/debian/copyright b/debian/copyright index b4bd7ea..8418831 100644 --- a/debian/copyright +++ b/debian/copyright @@ -1,11 +1,11 @@ This package was debianized by Kevin M. Rosenberg in October 2002. -The upstream source was downloaded from http://pubmed.b9.com/ +The upstream source was downloaded from http://entrez.b9.com/ Upstream Author: Kevin M. Rosenberg -Copyright (c) 2000-2002 Kevin M. Rosenberg +Copyright (c) 2000-2013 Kevin M. Rosenberg This code is free software; you can redistribute it and/or modify it under the terms of the version 2.1 of the GNU Lesser General Public diff --git a/debian/rules b/debian/rules index b98847e..a10399d 100755 --- a/debian/rules +++ b/debian/rules @@ -1,7 +1,7 @@ #!/usr/bin/make -f -pkg := pubmed -debpkg := cl-pubmed +pkg := entrez +debpkg := cl-entrez clc-source := usr/share/common-lisp/source clc-systems := usr/share/common-lisp/systems @@ -19,7 +19,7 @@ install: build dh_testroot dh_prep dh_installdirs - dh_install pubmed.asd $(clc-files) + dh_install entrez.asd $(clc-files) dh_install *.lisp $(clc-files) binary-indep: install diff --git a/debian/upload.sh b/debian/upload.sh index 49ec2ff..4be6929 100755 --- a/debian/upload.sh +++ b/debian/upload.sh @@ -1,4 +1,4 @@ #!/bin/bash -e -dup pubmed -Ufiles.med-info.com -D/home/ftp/pubmed -C"(umask 022; cd /opt/apache/htdocs/pubmed; make install)" -su $* +dup entrez -Ufiles.med-info.com -D/home/ftp/entrez -C"(umask 022; cd /opt/apache/htdocs/entrez; make install)" -su $* diff --git a/debian/watch b/debian/watch index 83f0bcf..b8109e0 100644 --- a/debian/watch +++ b/debian/watch @@ -1,2 +1,2 @@ version=3 -http://files.b9.com/pubmed/pubmed-([\d\.]*)\.tar\.gz +http://files.b9.com/entrez/entrez-([\d\.]*)\.tar\.gz diff --git a/entrez.asd b/entrez.asd new file mode 100644 index 0000000..727617a --- /dev/null +++ b/entrez.asd @@ -0,0 +1,36 @@ +;;;; -*- Mode: LISP; Syntax: ANSI-Common-Lisp; Base: 10 -*- +;;;; ************************************************************************* +;;;; FILE IDENTIFICATION +;;;; +;;;; Name: entrez.asd +;;;; Purpose: ASDF definition file for Entrez +;;;; Programmer: Kevin M. Rosenberg +;;;; Date Started: Sep 2002 +;;;; +;;;; This file, part of cl-entrez, is Copyright (c) 2002-2013 by Kevin M. Rosenberg +;;;; +;;;; cl-entrez users are granted the rights to distribute and use this software +;;;; as governed by the terms of the GNU Lesser General Public License +;;;; (http://www.gnu.org/licenses/lgpl.html) +;;;; ************************************************************************* + +(in-package #:cl-user) +(defpackage #:entrez-system (:use #:asdf #:cl)) +(in-package #:entrez-system) + +#+allegro (require :aserve) + +;; only define system on implementations that aserve is available +(defsystem entrez + :name "cl-entrez" + :author "Kevin M. Rosenberg " + :version "2.1" + :maintainer "Kevin M. Rosenberg " + :licence "GNU Lesser General Public License" + :description "Library for querying the Entrez medical literature database" + :long-description "This library has functions for querying the Entrez medical literature database and parsing the XML results into Common Lisp objects." + + :components ((:file "package") + (:file "entrez" :depends-on ("package"))) + + :depends-on (:kmrcl #-allegro :aserve)) diff --git a/entrez.lisp b/entrez.lisp new file mode 100644 index 0000000..89db4c7 --- /dev/null +++ b/entrez.lisp @@ -0,0 +1,376 @@ +;;;; -*- Mode: Lisp; Syntax: ANSI-Common-Lisp; Base: 10 -*- +;;;; ************************************************************************* +;;;; FILE IDENTIFICATION +;;;; +;;;; Name: entrez.lisp +;;;; Purpose: Library to access NCBI Entrez web application +;;;; Programmer: Kevin M. Rosenberg +;;;; Date Started: April 2013 +;;;; +;;;; This file, part of cl-entrez, is Copyright (c) 2002 by Kevin M. Rosenberg +;;;; +;;;; cl-entrez users are granted the rights to distribute and use this software +;;;; as governed by the terms of the GNU Lesser General Public License +;;;; (http://www.gnu.org/licenses/lgpl.html) +;;;; ************************************************************************* + +(in-package #:entrez) + + +(defparameter +entrez-host+ "eutils.ncbi.nlm.nih.gov") +(defparameter +entrez-search-url+ "/entrez/eutils/esearch.fcgi") +(defparameter +entrez-fetch-url+ "/entrez/eutils/efetch.fcgi") +(defparameter +entrez-summary-url+ "/entrez/eutils/esummary.fcgi") +(defparameter *proxy-host* nil) + + +(define-condition entrez-condition () + ()) + + +(define-condition entrez-server-error (error entrez-condition) + ((response :initarg :response + :initform nil + :reader entrez-condition-response)) + (:report (lambda (c stream) + (format stream "A Entrez server error occurred.") + (awhen (entrez-condition-response c) + (format stream " The server response was:~&~S" it))))) + +(define-condition entrez-query-error (error entrez-condition) + ((response :initarg :response + :initform nil + :reader entrez-condition-response)) + (:report (lambda (c stream) + (format stream "A Entrez server error occurred.") + (awhen (entrez-condition-response c) + (format stream " The server response was:~&~S" it))))) + +;;; Article-Set and Article Classes + +(defclass pm-article-set () + ((query :type string :initarg :query :accessor articles-query) + (webenv :type string :initarg :webenv :accessor articles-webenv) + (qkey :type string :initarg :qkey :accessor articles-qkey) + (articles :type list :initarg :articles :accessor articles) + (total :type fixnum :initarg :total :accessor articles-total) + (count :type fixnum :initarg :count :accessor articles-count) + (start :type fixnum :initarg :start :accessor articles-start)) + (:documentation "Entrez Article Set Class") + (:default-initargs :total 0 :start 0 :count 0 :query nil + :articles nil :qkey nil :webenv nil)) + +(defclass equery () + ((query :type string :initarg :query :accessor query) + (wenv :type string :initarg :webenv :accessor wenv) + (qkey :type string :initarg :qkey :accessor qkey) + (qcount :type fixnum :initarg :qcount :accessor qcount)) + (:documentation "Entrez EQuery Results Class") + (:default-initargs :qcount 0 :query nil :qkey nil :webenv nil)) + +(defclass pm-article () + ( + (pmid :type integer :accessor article-pmid) + (title :type string :accessor article-title) + (authors :type list :accessor article-authors) + (affiliation :type string :accessor article-affiliation) + (journal :type string :accessor article-journal) + (date :type string :accessor article-date) + (volume :type string :accessor article-volume) + (issue :type string :accessor article-issue) + (pages :type string :accessor article-pages) + (abstract :type string :accessor article-abstract) + (mesh-headings :type list :accessor article-mesh-headings)) + (:documentation "Entrez Article Class")) + +(defmethod print-object ((obj pm-article-set) (s stream)) + (print-unreadable-object (obj s :type t :identity t) + (format s "~d total articles, ~d articles starting at #~d" + (articles-total obj) + (articles-count obj) + (articles-start obj) + ))) + +(defmethod print-object ((obj equery) (s stream)) + (print-unreadable-object (obj s :type t :identity t) + (format s "Found ~d with qkey ~a & wenv ~a" + (qcount obj) + (qkey obj) + (wenv obj)))) + +(defmethod print-object ((obj pm-article) (s stream)) + (print-unreadable-object (obj s :type t :identity t) + (format s "pmid:~d, title:~S" (article-pmid obj) + (article-title obj)))) + +(defun article-equal-p (a b) + (check-type a pm-article) + (check-type b pm-article) + (eql (article-pmid a) (article-pmid b))) + +(defun article-ref (art) + "Return a string of publication data for an article" + (let ((ref "")) + (awhen (article-date art) + (string-append ref (format nil "~a; " it))) + (awhen (article-volume art) + (string-append ref it)) + (awhen (article-issue art) + (string-append ref (format nil "(~a)" it))) + (awhen (article-pages art) + (string-append ref (format nil ":~a" it))) + ref)) + +(defmethod print-article-set ((artset pm-article-set) + &key (os *standard-output*) (format :text) + (complete nil) (print-link nil)) + "Display an article set to specified stream in specified format" + (dotimes (i (articles-count artset) artset) + (if (nth i (articles artset)) + (print-article (nth i (articles artset)) :os os :format format + :complete complete :print-link print-link) + (princ "NULL Article" os)))) + +(defmethod print-article ((art pm-article) &key (os *standard-output*) + (format :text) (complete nil) (print-link nil)) + "Display an article" + (ecase format + (:text + (format os "~a~%~a~%~a~a ~a~%~a~%" + (article-title art) + (list-to-delimited-string (article-authors art) ", ") + (aif (article-affiliation art) + (format nil "~a~%" it) "") + (article-journal art) (article-ref art) + (aif (article-abstract art) + (if complete + it + "Abstract available") + "No abstract available") + (when complete + (format os "~a~%" (article-mesh-headings art))))) + (:html + (let ((has-link (or (article-abstract art) (article-mesh-headings art)))) + (when (and print-link has-link) + (format os "" (funcall print-link + (article-pmid art)))) + (format os "
~a
~%" + (article-title art)) + (when (and print-link has-link) + (format os "")) + (format os "
~a
~%" + (list-to-delimited-string (article-authors art) ", ")) + (format os "
~a ~a
~%" + (article-journal art) (article-ref art)) + (when (and complete (article-abstract art)) + (format os "
~a
~%" + (article-abstract art))) + (when (and complete (article-mesh-headings art)) + (format os "
Mesh Headings:
") + (dolist (mh (article-mesh-headings art)) + (format os "
~a
~%" mh))) + (format os "

~%")))) + art) + + +;;; Entrez Query Functions + +(defun equery (searchstr &key (db "pubmed") (usehistory t) start max) + "Performs Entrez query and returns article-set structure" + (multiple-value-bind + (xmlres ids) (do-equery searchstr :db db :usehistory usehistory :start start :max max) + (when (xml-tag-contents "Count" xmlres) + (let ((as (make-instance 'equery))) + (setf + (qcount as) (parse-integer (xml-tag-contents "Count" xmlres)) + (query as) searchstr) + (kmrcl:awhen (xml-tag-contents "WebEnv" xmlres) + (setf (wenv as) kmrcl:it)) + (kmrcl:awhen (xml-tag-contents "QueryKey" xmlres) + (setf (qkey as) kmrcl:it)) + as)))) + +(defun esummary (as &key maximum start) +;; (articles as) (extract-article-set results)) + as) + +(defun pm-fetch-ids (pmids) + "Fetchs list of Entrez ID's and returns pm-article-set class" + (setq pmids (mklist pmids)) + (let ((results (entrez-fetch-pmids-xml pmids))) + (unless (xml-tag-contents "Error" results) + (let ((as (make-instance 'pm-article-set))) + (setf + (articles-total as) (length pmids) + (articles-query as) (list-to-delimited-string pmids #\,) + (articles-start as) 0 + (articles-count as) (length pmids) + (articles as) (extract-article-set results)) + as)))) + +#+ignore +(defun entrez-search-tree (searchstr &key maximum start) + "Performs a entrez search and returns two values: +tree of Entrez search results and tree of Entrez search status" + (multiple-value-bind + (xml-search-results xml-search-status) + (entrez-search-xml searchstr :maximum maximum :start start) + (if xml-search-results + (values (parse-xml-no-ws xml-search-results) + (parse-xml-no-ws xml-search-status)) + (values nil (parse-xml-no-ws xml-search-status))))) + +(defun do-equery (term &key (db "pubmed") start max (usehistory t)) + "Performs a Entrez search and returns two values: +XML string of Entrez search results and XML search status" + (let ((res (http-equery term :start start :max max + :usehistory usehistory :db db))) + (values res (extract-pmid-list res)))) + +(defun http-equery (term &key start max (usehistory t) + (db "pubmed")) + "Performs a Entrez search and returns XML results of Entrez search + which contains Entrez ID's and status results" + (let* ((alist `(("db" . ,db) ("term" . ,term))) + (geturl (format nil "http://~a~a" +entrez-host+ +entrez-search-url+))) + (when usehistory + (setq alist (acons "usehistory" "y" alist))) + (when start + (setq alist (acons "retstart" start alist))) + (when max + (setq alist (acons "retmax" max alist))) + (multiple-value-bind (xmlres http-code http-server-alist) + (net.aserve.client:do-http-request + geturl + :method :get + :query alist + :proxy *proxy-host*) + (declare (ignore http-server-alist)) + (when (not (equal http-code 200)) + (error "HTTP code performing '~A' was ~D." geturl http-code)) + xmlres))) + +(defun entrez-fetch-pmids-xml (pmids &key (db "pubmed")) + "Fetch articles for a list of Entrez ID's and return XML string" + (setq pmids (mklist pmids)) ;; Ensure list + (when pmids + (net.aserve.client:do-http-request + (format nil "http://~a~a" +entrez-host+ +entrez-fetch-url+) + :method :get + :query + `(("db" . ,db) ("report" . "xml") ("mode" . "text") + ("id" . ,(list-to-delimited-string pmids #\,))) + :proxy *proxy-host*))) + +;;; XML Extraction Routines + +(defun extract-article-set (results) + "Extract article set from Entrez XML string, return results in pm-article-set class" + (multiple-value-bind (as-start as-end as-next) + (positions-xml-tag-contents "PubmedArticleSet" results) + (declare (ignore as-end as-next)) + (when as-start + (let ((done nil) + (articles '()) + (pos as-start)) + (until done + (multiple-value-bind + (a-start a-end a-next) + (positions-xml-tag-contents "PubmedArticle" results pos) + (if a-start + (progn + (push (extract-article results a-start a-end) articles) + (setq pos a-next) + ) + (setq done t)))) + (nreverse articles))))) + +(defun extract-article (xmlstr a-start a-end) + "Extract article contents from Entrez XML string and return results in pm-article class" + (let ((article (make-instance 'pm-article))) + (setf + (article-pmid article) (parse-integer (xml-tag-contents "PMID" xmlstr a-start a-end)) + (article-title article) (xml-tag-contents "ArticleTitle" xmlstr a-start a-end) + (article-journal article) (xml-tag-contents "MedlineTA" xmlstr a-start a-end) + (article-pages article) (xml-tag-contents "MedlinePgn" xmlstr a-start a-end) + (article-affiliation article) (xml-tag-contents "Affiliation" xmlstr a-start a-end) + (article-abstract article) (xml-tag-contents "AbstractText" xmlstr a-start a-end)) + (multiple-value-bind (ji-start ji-end ji-next) + (positions-xml-tag-contents "JournalIssue" xmlstr a-start a-end) + (declare (ignore ji-next)) + (setf + (article-volume article) (xml-tag-contents "Volume" xmlstr ji-start ji-end) + (article-issue article) (xml-tag-contents "Issue" xmlstr ji-start ji-end)) + (aif (xml-tag-contents "MedlineDate" xmlstr ji-start ji-end) + (setf (article-date article) it) + (setf (article-date article) + (concatenate 'string (xml-tag-contents "Year" xmlstr ji-start ji-end) + (aif (xml-tag-contents "Month" xmlstr ji-start ji-end) + (format nil " ~a" it) + ""))))) + + (multiple-value-bind (al-start al-end al-next) + (positions-xml-tag-contents "AuthorList" xmlstr a-start a-end) + (declare (ignore al-next)) + (setf (article-authors article) + (when al-start + (let ((done nil) + (authors '()) + (pos al-start)) + (until done + (multiple-value-bind + (au-start au-end au-next) + (positions-xml-tag-contents "Author" xmlstr pos al-end) + (if au-start + (progn + (push (extract-author xmlstr au-start au-end) authors) + (setq pos au-next)) + (setq done t)))) + (nreverse authors))))) + + (multiple-value-bind (mhl-start mhl-end mhl-next) + (positions-xml-tag-contents "MeshHeadingList" xmlstr a-start a-end) + (declare (ignore mhl-next)) + (setf (article-mesh-headings article) + (when mhl-start + (let ((done nil) + (mesh-headings '()) + (pos mhl-start)) + (until done + (multiple-value-bind + (mh-start mh-end mh-next) + (positions-xml-tag-contents "MeshHeading" xmlstr pos mhl-end) + (if mh-start + (progn + (push (extract-mesh-heading xmlstr mh-start mh-end) mesh-headings) + (setq pos mh-next) + ) + (setq done t)))) + (nreverse mesh-headings))))) + + article)) + +(defun extract-author (xmlstr start end) + "Extract author name from XML string" + (let ((last-name (xml-tag-contents "LastName" xmlstr start end)) + (initials (xml-tag-contents "Initials" xmlstr start end))) + (concatenate 'string last-name " " initials))) + +(defun extract-mesh-heading (xmlstr start end) + "Extract and format mesh headings from XML string" + (let ((desc (xml-tag-contents "DescriptorName" xmlstr start end)) + (sh (xml-tag-contents "SubHeading" xmlstr start end))) + (if sh + (format nil "~a(~a)" desc sh) + desc))) + +(defun extract-pmid-list (results) + "Returns list of Entrez ID's from XML result string" + (cond + ((search "" results) + (error 'entrez-query-error :response results)) + ((search "

Server Error

" results) + (error 'entrez-server-error :response results)) + (t + (awhen (xml-tag-contents "Id" results) + (delimited-string-to-list it #\space))))) diff --git a/package.lisp b/package.lisp index b7bfe49..9bb14c6 100644 --- a/package.lisp +++ b/package.lisp @@ -3,13 +3,11 @@ ;;;; FILE IDENTIFICATION ;;;; ;;;; Name: package.lisp -;;;; Purpose: Package file for cl-pubmed +;;;; Purpose: Package file for cl-entrez ;;;; Programmer: Kevin M. Rosenberg ;;;; Date Started: Jun 2001 ;;;; -;;;; $Id$ -;;;; -;;;; This file, part of cl-pubmed, is Copyright (c) 2002 by Kevin M. Rosenberg +;;;; This file, part of cl-pubmed, is Copyright (c) 2002-2013 by Kevin M. Rosenberg ;;;; ;;;; cl-pubmed users are granted the rights to distribute and use this software ;;;; as governed by the terms of the GNU Lesser General Public License @@ -18,18 +16,20 @@ (in-package #:cl-user) -(defpackage #:pubmed +(defpackage #:entrez (:use #:common-lisp #:kmrcl) (:export ;; Conditions - #:pubmed-condition - #:pubmed-query-error - #:pubmed-server-error - #:pubmed-condition-response + #:entrez-condition + #:entrez-query-error + #:entrez-server-error + #:entrez-condition-response ;; Query functions - #:pm-query + #:equery + #:esummary + #:efetch #:pm-fetch-ids ;; Print functions diff --git a/pubmed-src.lisp b/pubmed-src.lisp deleted file mode 100644 index ad18ca9..0000000 --- a/pubmed-src.lisp +++ /dev/null @@ -1,355 +0,0 @@ -;;;; -*- Mode: Lisp; Syntax: ANSI-Common-Lisp; Base: 10 -*- -;;;; ************************************************************************* -;;;; FILE IDENTIFICATION -;;;; -;;;; Name: pubmed-src.lisp -;;;; Purpose: Library to access PubMed web application -;;;; Programmer: Kevin M. Rosenberg -;;;; Date Started: Jun 2001 -;;;; -;;;; $Id$ -;;;; -;;;; This file, part of cl-pubmed, is Copyright (c) 2002 by Kevin M. Rosenberg -;;;; -;;;; cl-pubmed users are granted the rights to distribute and use this software -;;;; as governed by the terms of the GNU Lesser General Public License -;;;; (http://www.gnu.org/licenses/lgpl.html) -;;;; ************************************************************************* - -(in-package #:pubmed) - - -(defparameter +pubmed-host+ "www.ncbi.nlm.nih.gov") -(defparameter +pubmed-query-url+ "/entrez/utils/pmqty.fcgi") -(defparameter +pubmed-fetch-url+ "/entrez/utils/pmfetch.fcgi") -(defparameter *proxy-host* nil) - - -(define-condition pubmed-condition () - ()) - - -(define-condition pubmed-server-error (error pubmed-condition) - ((response :initarg :response - :initform nil - :reader pubmed-condition-response)) - (:report (lambda (c stream) - (format stream "A PubMed server error occurred.") - (awhen (pubmed-condition-response c) - (format stream " The server response was:~&~S" it))))) - -(define-condition pubmed-query-error (error pubmed-condition) - ((response :initarg :response - :initform nil - :reader pubmed-condition-response)) - (:report (lambda (c stream) - (format stream "A PubMed server error occurred.") - (awhen (pubmed-condition-response c) - (format stream " The server response was:~&~S" it))))) - -;;; Article-Set and Article Classes - -(defclass pm-article-set () - ((query :type string :initarg :query :accessor articles-query) - (articles :type list :initarg :articles :accessor articles) - (total :type fixnum :initarg :total :accessor articles-total) - (count :type fixnum :initarg :count :accessor articles-count) - (start :type fixnum :initarg :start :accessor articles-start)) - (:documentation "Pubmed Article Set Class") - (:default-initargs :total 0 :start 0 :count 0 - :query nil :articles nil)) - -(defclass pm-article () - ( - (pmid :type integer :accessor article-pmid) - (title :type string :accessor article-title) - (authors :type list :accessor article-authors) - (affiliation :type string :accessor article-affiliation) - (journal :type string :accessor article-journal) - (date :type string :accessor article-date) - (volume :type string :accessor article-volume) - (issue :type string :accessor article-issue) - (pages :type string :accessor article-pages) - (abstract :type string :accessor article-abstract) - (mesh-headings :type list :accessor article-mesh-headings)) - (:documentation "Pubmed Article Class")) - -(defmethod print-object ((obj pm-article-set) (s stream)) - (print-unreadable-object (obj s :type t :identity t) - (format s "~d total articles, ~d articles starting at #~d" - (articles-total obj) - (articles-count obj) - (articles-start obj) - ))) - -(defmethod print-object ((obj pm-article) (s stream)) - (print-unreadable-object (obj s :type t :identity t) - (format s "pmid:~d, title:~S" (article-pmid obj) - (article-title obj)))) - -(defun article-equal-p (a b) - (check-type a pm-article) - (check-type b pm-article) - (eql (article-pmid a) (article-pmid b))) - -(defun article-ref (art) - "Return a string of publication data for an article" - (let ((ref "")) - (awhen (article-date art) - (string-append ref (format nil "~a; " it))) - (awhen (article-volume art) - (string-append ref it)) - (awhen (article-issue art) - (string-append ref (format nil "(~a)" it))) - (awhen (article-pages art) - (string-append ref (format nil ":~a" it))) - ref)) - -(defmethod print-article-set ((artset pm-article-set) - &key (os *standard-output*) (format :text) - (complete nil) (print-link nil)) - "Display an article set to specified stream in specified format" - (dotimes (i (articles-count artset) artset) - (if (nth i (articles artset)) - (print-article (nth i (articles artset)) :os os :format format - :complete complete :print-link print-link) - (princ "NULL Article" os)))) - -(defmethod print-article ((art pm-article) &key (os *standard-output*) - (format :text) (complete nil) (print-link nil)) - "Display an article" - (ecase format - (:text - (format os "~a~%~a~%~a~a ~a~%~a~%" - (article-title art) - (list-to-delimited-string (article-authors art) ", ") - (aif (article-affiliation art) - (format nil "~a~%" it) "") - (article-journal art) (article-ref art) - (aif (article-abstract art) - (if complete - it - "Abstract available") - "No abstract available") - (when complete - (format os "~a~%" (article-mesh-headings art))))) - (:html - (let ((has-link (or (article-abstract art) (article-mesh-headings art)))) - (when (and print-link has-link) - (format os "" (funcall print-link - (article-pmid art)))) - (format os "
~a
~%" - (article-title art)) - (when (and print-link has-link) - (format os "")) - (format os "
~a
~%" - (list-to-delimited-string (article-authors art) ", ")) - (format os "
~a ~a
~%" - (article-journal art) (article-ref art)) - (when (and complete (article-abstract art)) - (format os "
~a
~%" - (article-abstract art))) - (when (and complete (article-mesh-headings art)) - (format os "
Mesh Headings:
") - (dolist (mh (article-mesh-headings art)) - (format os "
~a
~%" mh))) - (format os "

~%")))) - art) - - -;;; PubMed Query Functions - -(defun pm-query (searchstr &key maximum start) - "Performs PubMed query and fetch and returns article-set structure" - (multiple-value-bind - (results status) - (pubmed-search-xml searchstr :maximum maximum :start start) - (when (xml-tag-contents "Count" status) - (let ((as (make-instance 'pm-article-set))) - (setf - (articles-total as) (parse-integer (xml-tag-contents "Count" status)) - (articles-query as) searchstr - (articles-start as) (parse-integer (xml-tag-contents "DispStart" status)) - (articles-count as) (parse-integer (xml-tag-contents "DispMax" status)) - (articles as) (extract-article-set results)) - as)))) - -(defun pm-fetch-ids (pmids) - "Fetchs list of Pubmed ID's and returns pm-article-set class" - (setq pmids (mklist pmids)) - (let ((results (pubmed-fetch-pmids-xml pmids))) - (unless (xml-tag-contents "Error" results) - (let ((as (make-instance 'pm-article-set))) - (setf - (articles-total as) (length pmids) - (articles-query as) (list-to-delimited-string pmids #\,) - (articles-start as) 0 - (articles-count as) (length pmids) - (articles as) (extract-article-set results)) - as)))) - -#+ignore -(defun pubmed-search-tree (searchstr &key maximum start) - "Performs a pubmed search and returns two values: -tree of PubMed search results and tree of PubMed search status" - (multiple-value-bind - (xml-search-results xml-search-status) - (pubmed-search-xml searchstr :maximum maximum :start start) - (if xml-search-results - (values (parse-xml-no-ws xml-search-results) - (parse-xml-no-ws xml-search-status)) - (values nil (parse-xml-no-ws xml-search-status))))) - -(defun pubmed-search-xml (searchstr &key maximum start) - "Performs a Pubmed search and returns two values: -XML string of PubMed search results and XML search status" - (multiple-value-bind - (pmids search-status) - (pubmed-query-xml searchstr :maximum maximum :start start) - (values (pubmed-fetch-pmids-xml pmids) search-status))) - -(defun pubmed-query-xml (searchstr &key maximum start) - "Performs a Pubmed search and returns two values: - list of PubMed ID's that match search string and XML search status" - (let ((search-results (pubmed-query-status searchstr :maximum maximum :start start))) - (values (extract-pmid-list search-results) search-results))) - -(defun pubmed-query-status (searchstr &key start maximum) - "Performs a Pubmed search and returns XML results of PubMed search - which contains PubMed ID's and status results" - (let ((query-alist `(("db" . "m") ("term" . ,searchstr) ("mode" . "xml")))) - (when maximum (push (cons "dispmax" maximum) query-alist)) - (when start (push (cons "dispstart" start) query-alist)) - (net.aserve.client:do-http-request - (format nil "http://~a~a" +pubmed-host+ +pubmed-query-url+) - :method :get - :query query-alist - :proxy *proxy-host*))) - -(defun pubmed-fetch-pmids-xml (pmids) - "Fetch articles for a list of PubMed ID's and return XML string" - (setq pmids (mklist pmids)) ;; Ensure list - (when pmids - (net.aserve.client:do-http-request - (format nil "http://~a~a" +pubmed-host+ +pubmed-fetch-url+) - :method :get - :query - `(("db" . "PubMed") ("report" . "xml") ("mode" . "text") - ("id" . ,(list-to-delimited-string pmids #\,))) - :proxy *proxy-host*))) - -;;; XML Extraction Routines - -(defun extract-article-set (results) - "Extract article set from PubMed XML string, return results in pm-article-set class" - (multiple-value-bind (as-start as-end as-next) - (positions-xml-tag-contents "PubmedArticleSet" results) - (declare (ignore as-end as-next)) - (when as-start - (let ((done nil) - (articles '()) - (pos as-start)) - (until done - (multiple-value-bind - (a-start a-end a-next) - (positions-xml-tag-contents "PubmedArticle" results pos) - (if a-start - (progn - (push (extract-article results a-start a-end) articles) - (setq pos a-next) - ) - (setq done t)))) - (nreverse articles))))) - -(defun extract-article (xmlstr a-start a-end) - "Extract article contents from PubMed XML string and return results in pm-article class" - (let ((article (make-instance 'pm-article))) - (setf - (article-pmid article) (parse-integer (xml-tag-contents "PMID" xmlstr a-start a-end)) - (article-title article) (xml-tag-contents "ArticleTitle" xmlstr a-start a-end) - (article-journal article) (xml-tag-contents "MedlineTA" xmlstr a-start a-end) - (article-pages article) (xml-tag-contents "MedlinePgn" xmlstr a-start a-end) - (article-affiliation article) (xml-tag-contents "Affiliation" xmlstr a-start a-end) - (article-abstract article) (xml-tag-contents "AbstractText" xmlstr a-start a-end)) - (multiple-value-bind (ji-start ji-end ji-next) - (positions-xml-tag-contents "JournalIssue" xmlstr a-start a-end) - (declare (ignore ji-next)) - (setf - (article-volume article) (xml-tag-contents "Volume" xmlstr ji-start ji-end) - (article-issue article) (xml-tag-contents "Issue" xmlstr ji-start ji-end)) - (aif (xml-tag-contents "MedlineDate" xmlstr ji-start ji-end) - (setf (article-date article) it) - (setf (article-date article) - (concatenate 'string (xml-tag-contents "Year" xmlstr ji-start ji-end) - (aif (xml-tag-contents "Month" xmlstr ji-start ji-end) - (format nil " ~a" it) - ""))))) - - (multiple-value-bind (al-start al-end al-next) - (positions-xml-tag-contents "AuthorList" xmlstr a-start a-end) - (declare (ignore al-next)) - (setf (article-authors article) - (when al-start - (let ((done nil) - (authors '()) - (pos al-start)) - (until done - (multiple-value-bind - (au-start au-end au-next) - (positions-xml-tag-contents "Author" xmlstr pos al-end) - (if au-start - (progn - (push (extract-author xmlstr au-start au-end) authors) - (setq pos au-next)) - (setq done t)))) - (nreverse authors))))) - - (multiple-value-bind (mhl-start mhl-end mhl-next) - (positions-xml-tag-contents "MeshHeadingList" xmlstr a-start a-end) - (declare (ignore mhl-next)) - (setf (article-mesh-headings article) - (when mhl-start - (let ((done nil) - (mesh-headings '()) - (pos mhl-start)) - (until done - (multiple-value-bind - (mh-start mh-end mh-next) - (positions-xml-tag-contents "MeshHeading" xmlstr pos mhl-end) - (if mh-start - (progn - (push (extract-mesh-heading xmlstr mh-start mh-end) mesh-headings) - (setq pos mh-next) - ) - (setq done t)))) - (nreverse mesh-headings))))) - - article)) - -(defun extract-author (xmlstr start end) - "Extract author name from XML string" - (let ((last-name (xml-tag-contents "LastName" xmlstr start end)) - (initials (xml-tag-contents "Initials" xmlstr start end))) - (concatenate 'string last-name " " initials))) - -(defun extract-mesh-heading (xmlstr start end) - "Extract and format mesh headings from XML string" - (let ((desc (xml-tag-contents "DescriptorName" xmlstr start end)) - (sh (xml-tag-contents "SubHeading" xmlstr start end))) - (if sh - (format nil "~a(~a)" desc sh) - desc))) - -(defun extract-pmid-list (results) - "Returns list of PubMed ID's from XML result string" - (cond - ((search "" results) - (error 'pubmed-query-error :response results)) - ((search "

Server Error

" results) - (error 'pubmed-server-error :response results)) - (t - (awhen (xml-tag-contents "Id" results) - (delimited-string-to-list it #\space))))) - - - diff --git a/pubmed.asd b/pubmed.asd deleted file mode 100644 index b794be8..0000000 --- a/pubmed.asd +++ /dev/null @@ -1,43 +0,0 @@ -;;;; -*- Mode: LISP; Syntax: ANSI-Common-Lisp; Base: 10 -*- -;;;; ************************************************************************* -;;;; FILE IDENTIFICATION -;;;; -;;;; Name: pubmed.asd -;;;; Purpose: ASDF definition file for Pubmed -;;;; Programmer: Kevin M. Rosenberg -;;;; Date Started: Sep 2002 -;;;; -;;;; $Id$ -;;;; -;;;; This file, part of cl-pubmed, is Copyright (c) 2002 by Kevin M. Rosenberg -;;;; -;;;; cl-pubmed users are granted the rights to distribute and use this software -;;;; as governed by the terms of the GNU Lesser General Public License -;;;; (http://www.gnu.org/licenses/lgpl.html) -;;;; ************************************************************************* - -(in-package #:cl-user) -(defpackage #:pubmed-system (:use #:asdf #:cl)) -(in-package #:pubmed-system) - - -#+(and allegro common-lisp-controller) (c-l-c::clc-require :aserve) -#+(and allegro (not common-lisp-controller)) (require :aserve) - -;; only define system on implementations that aserve is available -(defsystem pubmed - :name "cl-pubmed" - :author "Kevin M. Rosenberg " - :version "2.1" - :maintainer "Kevin M. Rosenberg " - :licence "GNU Lesser General Public License" - :description "Library for querying the PubMed medical literature database" - :long-description "This library has functions for querying the PubMed medical literature database and parsing the XML results into Common Lisp objects." - - :components ((:file "package") - (:file "pubmed-src" :depends-on ("package"))) - - :depends-on (:kmrcl #-allegro :aserve)) - - -