From 6c686e0631823d747ce59de4bcd29513f9b30295 Mon Sep 17 00:00:00 2001
From: Munyoki Kilyungi
Date: Thu, 8 Feb 2024 17:13:27 +0300
Subject: Add a basic guile script to transform stored rtf data to files.

Signed-off-by: Munyoki Kilyungi <me@bonfacemunyoki.com>
---
 examples/dataset-metadata-git.scm | 76 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 76 insertions(+)
 create mode 100755 examples/dataset-metadata-git.scm

diff --git a/examples/dataset-metadata-git.scm b/examples/dataset-metadata-git.scm
new file mode 100755
index 0000000..f2bde48
--- /dev/null
+++ b/examples/dataset-metadata-git.scm
@@ -0,0 +1,76 @@
+#! /usr/bin/env guile
+
+!#
+(use-modules
+ (ice-9 getopt-long)
+ (srfi srfi-26)
+ ((ice-9 regex) #:select (regexp-substitute/global))
+ ((transform strings) #:select (string-blank?))
+ ((transform sql) #:select (call-with-target-database sql-for-each))
+ ((transform triples) #:select (string->identifier)))
+
+(define (save-file file result)
+  (when result
+    (let ((dir-name (dirname file)))
+      (unless (file-exists? dir-name)
+        (mkdir dir-name))
+      (with-output-to-file file
+        (lambda ()
+          (format #t result))))))
+
+(define (infopages/sql->rtf result)
+  (let* ((get (cut assoc-ref result <>))
+         (get* (compose (lambda (str)
+                           (if (string-blank? str)
+                               #f
+                               str))
+                         get))
+         (identifier (string->identifier
+                      ""
+                      (regexp-substitute/global #f "[^A-Za-z0-9:]"
+                                                (get "InfoPageName")
+                                                'pre "_" 'post)))
+         (dir-name (string-append "/export/data/genenetwork/data/"))
+         (file-name (cut string-append dir-name <>))
+         (summary (get* "Summary"))
+         (tissue (get* "AboutTissue"))
+         (specifics (get* "Specifics"))
+         (cases (get* "AboutCases"))
+         (platform (get* "AboutPlatform"))
+         (processing (get* "AboutDataProcessing"))
+         (notes (get* "Notes"))
+         (experiment-design (get* "ExperimentDesign"))
+         (acknowledgment (get* "Acknowledgment")))
+    (for-each (lambda (x)
+                (save-file (file-name (string-append dir-name (car x)))
+                           (cdr x)))
+              `(("summary.rtf" . ,summary)
+                ("tissue.rtf" . ,tissue)
+                ("specifics.rtf" . ,specifics)
+                ("cases.rtf" . ,cases)
+                ("platform.rtf" . ,platform)
+                ("processing.rtf" . ,processing)
+                ("notes.rtf" . ,notes)
+                ("experiment-design.rtf" . ,experiment-design)
+                ("acknowledgement.rtf" . ,acknowledgement)))))
+
+(let* ((option-spec
+        '((settings (single-char #\s) (value #t))
+          (output (single-char #\o) (value #t))))
+       (options (getopt-long (command-line) option-spec))
+       (settings (option-ref options 'settings #f))
+       (output (option-ref options 'output #f))
+       (query "SELECT InfoPageName, Datasets.Summary, Datasets.AboutTissue, InfoFiles.Specifics,
+Datasets.AboutCases, Datasets.AboutPlatform, Datasets.AboutDataProcessing,
+Datasets.Notes, Datasets.ExperimentDesign, Datasets.Acknowledgment
+FROM InfoFiles LEFT JOIN Datasets USING (DatasetId) LIMIT 2")
+       (%connection-settings
+        (call-with-input-file settings
+          read)))
+  ;; TODO
+  (call-with-target-database
+   %connection-settings
+   (lambda (db)
+     (sql-for-each infopages/sql->rtf
+                   db
+                   query))))
-- 
cgit v1.2.3