From 3b19dc04b14e27a12a3152d0cc3a52d1048854a8 Mon Sep 17 00:00:00 2001
From: Munyoki Kilyungi
Date: Fri, 29 Jul 2022 15:57:34 +0300
Subject: Add new functions to read CSV files in a flexible way.

We add new functions---parse-raw-table and parse-aggregate-table---to
parse raw ungrouped data and grouped data respectively.

* csv-dump.scm (assoc-set, parse-raw-table, parse-aggregate-table):
New functions.

Signed-off-by: Arun Isaac <arunisaac@systemreboot.net>
---
 csv-dump.scm | 66 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 66 insertions(+)
 create mode 100644 csv-dump.scm

diff --git a/csv-dump.scm b/csv-dump.scm
new file mode 100644
index 0000000..a2a9127
--- /dev/null
+++ b/csv-dump.scm
@@ -0,0 +1,66 @@
+(use-modules (srfi srfi-1)
+             (ice-9 match)
+             (ice-9 rdelim)
+             (dsv))
+
+(define (assoc-set alist key value)
+  "Associate KEY with VALUE in ALIST, and return the new association
+list. This is a functional setter and does not mutate ALIST. KEY is
+compared with keys in ALIST using equal?."
+  (acons key value
+         (alist-delete key alist)))
+
+(define (parse-aggregate-table table)
+  "Convert TABLE from the provided aggregate file to an s-exp equivalent.
+The GROUP forms the KEY of each alist."
+  (match table
+    ((header table-body ...)
+     (fold (lambda (row results)
+             (let ((alist (map cons header row)))
+               (acons (string->symbol (assoc-ref alist "group"))
+                      (filter-map (lambda (key value)
+                                    (and (member key
+                                                 (list "mean" "n" "sex " "SD" "day"))
+                                         (cons (string->symbol (string-downcase key))
+                                               value)))
+                                  header row)
+                      results)))
+           '()
+           table-body))))
+
+(define (parse-raw-table table aggregate-table)
+  "Given TABLE which represents the data from the raw file, and the
+AGGREGATE-TABLE which is an ALIST containing aggregate data, add extra
+metadata not present in the aggregate table and return this result."
+  (match table
+    ((header table-body ...)
+     (fold (lambda (row results)
+             (let* ((alist (map cons header row))
+                    (key (string->symbol (string-join
+                                          (list (assoc-ref alist "strain")
+                                                (assoc-ref alist "sex")
+                                                (assoc-ref alist "day"))
+                                          "_")))
+                    (aggregate-data (assq-ref aggregate-table key))
+                    (folded-data (assq-ref results key)))
+               ;; Return an updated version of the aggregate-table as
+               ;; you loop through the raw table.  Do this while
+               ;; collecting the values for 'bw and setting 'inf-dose
+               (assoc-set results
+                          key (assoc-set
+                               ;; inf_dose is repetitive in raw table
+                               ;; and it's common in the same vector
+                               ;; of data.  If it isn't set, set it
+                               ;; once.
+                               (if (assq-ref folded-data 'aggregate-dose)
+                                   aggregate-data
+                                   (acons 'inf-dose (assoc-ref alist "inf_dose") aggregate-data))
+                               'bw
+                               ;; When looping in the raw table,
+                               ;; collect all the 'bw values.
+                               (if (assq-ref folded-data 'bw)
+                                   (cons (assoc-ref alist "BW")
+                                         (assq-ref folded-data 'bw))
+                                   (list (assoc-ref alist "BW")))))))
+           '()
+           table-body))))
-- 
cgit v1.2.3