library( irs990efile ) # functions for parsing efile XMLs
library( dplyr ) # data wrangling
library( purrr ) # data wrangling
library( pander ) # formatting
library( knitr ) # formatting
library( DT ) # table printing
library( httr ) # web requests
library( xmltools ) # xml utilities
library( xml2 ) # xml utilities
library( XML ) # xml utilities
library( data.tree ) # network visualization
library( networkD3 ) # network visualization
library( igraph ) # network visualization
source( "R/rdb-functions-v2.R")
source( "R/utils.R")
head( index )
The concordance is an RDB to XML Crosswalk that maps all xpaths onto a specified set of variables. It is located in the irs990efile package.
The full data dictionary is available on GitHub.
%>%
concordance filter( rdb_table == table.name )
xpath | id |
---|---|
/Return/ReturnData/IRS990ScheduleA/Form990ScheduleAPartI/SupportedOrgInformation/OrganizedInUS | 1 |
/Return/ReturnData/IRS990ScheduleA/Form990ScheduleAPartI/SupportedOrgInformation/SupportedOrganizationNotified | 2 |
/Return/ReturnData/IRS990ScheduleA/SupportedOrgInformation/OrganizedInUS | 3 |
/Return/ReturnData/IRS990ScheduleA/SupportedOrgInformation/SupportedOrganizationNotified | 4 |
/Return/ReturnData/IRS990ScheduleA/SupportedOrgInformationGrp/SupportedOrgNotifiedInd | 5 |
/Return/ReturnData/IRS990ScheduleA/SupportedOrgInformationGrp/USOrganizedInd | 6 |
<- as.character( t.xpaths$xpath )
xpaths <- create_edgelist_v1( xpaths )
el <- FromDataFrameNetwork( network=el )
nd print( nd )
## levelName
## 1 Return
## 2 °--ReturnData
## 3 °--IRS990ScheduleA
## 4 ¦--Form990ScheduleAPartI
## 5 ¦ °--SupportedOrgInformation
## 6 ¦ ¦--OrganizedInUS
## 7 ¦ °--SupportedOrganizationNotified
## 8 ¦--SupportedOrgInformation
## 9 ¦ ¦--OrganizedInUS
## 10 ¦ °--SupportedOrganizationNotified
## 11 °--SupportedOrgInformationGrp
## 12 ¦--SupportedOrgNotifiedInd
## 13 °--USOrganizedInd
SetGraphStyle( nd, rankdir = "LR")
SetEdgeStyle( nd, arrowhead = "vee", color = "grey20", penwidth = 2 )
SetNodeStyle( nd,
style = "filled,rounded",
shape = "box",
fillcolor = "LightBlue",
fontname = "helvetica",
fontcolor="black",
tooltip = GetDefaultTooltip )
# SetNodeStyle(acme$IT, fillcolor = "LightBlue", penwidth = "5px")
plot( nd )
# old version
find_group_names( table.name )
## [1] "//SupportedOrgInformationGrp" "//Form990ScheduleAPartI"
## [3] "//SupportedOrgInformation"
# new version
find_table_headers( table.name )
## [1] "//SupportedOrgInformationGrp/SupportedOrganizationName"
## [2] "//SupportedOrgInformationGrp/EIN"
## [3] "//SupportedOrgInformationGrp/OrganizationTypeCd"
## [4] "//SupportedOrgInformationGrp/OrganizationTypeDesc"
## [5] "//SupportedOrgInformationGrp/GoverningDocumentListedInd"
## [6] "//SupportedOrgInformationGrp/SupportAmt"
## [7] "//SupportedOrgInformationGrp/OtherSupportAmt"
## [8] "//SupportedOrgInformationGrp/SupportedOrgNotifiedInd"
## [9] "//SupportedOrgInformationGrp/USOrganizedInd"
## [10] "//Form990ScheduleAPartI/Church170b1Ai"
## [11] "//Form990ScheduleAPartI/School170b1Aii"
## [12] "//Form990ScheduleAPartI/Hospital170b1Aiii"
## [13] "//Form990ScheduleAPartI/HospitalNameAndAddress"
## [14] "//Form990ScheduleAPartI/MedicalResearchOrg170b1Aiii"
## [15] "//Form990ScheduleAPartI/CollegeSupportOrg170b1Aiv"
## [16] "//Form990ScheduleAPartI/GovernmentalUnit170b1Av"
## [17] "//Form990ScheduleAPartI/PubliclySupportedOrg170b1Avi"
## [18] "//Form990ScheduleAPartI/CommunityTrust170b1Avi"
## [19] "//Form990ScheduleAPartI/PubliclySupportedOrg509a2"
## [20] "//Form990ScheduleAPartI/TestingForPublicSafety509a4"
## [21] "//Form990ScheduleAPartI/SupportingOrganization509a3"
## [22] "//Form990ScheduleAPartI/SupportingOrg509a3Type1"
## [23] "//Form990ScheduleAPartI/SupportingOrg509a3Type2"
## [24] "//Form990ScheduleAPartI/SupportingOrg509a3Type3FuncInt"
## [25] "//Form990ScheduleAPartI/WrittenIRSDetermination"
## [26] "//Form990ScheduleAPartI/SupportedOrgInformation"
## [27] "//Form990ScheduleAPartI/SumOfAmounts"
## [28] "//Form990ScheduleAPartI/SupportingOrg509a3Type3Other"
## [29] "//Form990ScheduleAPartI/CertificationCheckbox"
## [30] "//Form990ScheduleAPartI/ContribByControllingIndividual"
## [31] "//Form990ScheduleAPartI/ContributionByRelatedParty"
## [32] "//Form990ScheduleAPartI/ContribBy35ControlledEntity"
## [33] "//SupportedOrgInformation/Form990ScheduleAPartI"
## [34] "//SupportedOrgInformation/SupportedOrgInformation"
# manual
if( ! is.null(table.headers) )
{table.headers}
# create new folder
dir.create( folder.name )
setwd( folder.name )
# create sample
set.seed( 1234 )
<- sample_n( index, sample.size )
sample.index write.csv( index, "sample-index.csv", row.names=F )
<- sample.index$URL
sample.urls
# erase existing log files
file.create("XPATH-LOG.txt")
file.create("FAIL-LOG.txt")
<- Sys.time() # --------------------
start.build.time
<- list()
results.list <- list()
xpath.list
for( i in 1:length( sample.urls ) )
{
<- sample.urls[i]
url
<-
results.list[[i]] build_rdb_table_v2( url,
table.name, table.headers=table.headers )
<-
xpath.list[[i]] get_table_xpaths( url,
table.name, table.headers=table.headers )
if( i %% 100 == 0 ){ print(i) }
}
<- Sys.time() # --------------------
end.build.time
<- dplyr::bind_rows( results.list )
df
<- readLines( "FAIL-LOG.txt" )
fail.log <- xpath.list %>% unlist() %>% unique() %>% sort()
all.xpaths cat( all.xpaths, sep="\n", file="XPATH-LOG.txt" )
TABLE: SA-P99-T01-PUBLIC-CHARITY-STATUS
## [1] "NUMBER OF RETURNS PROCESSED: 1000"
## [1] "TOTAL NUMBER OF TABLE ROWS: 90"
TOTAL RUN TIME:
## Time difference of 16.65711 mins
## Time difference of -0.2776185 hours
write.csv( df, paste0( folder.name, "/", table.name, ".csv" ), row.names=F )
head(df)
Are all names standardized?
names(df)
## [1] "OBJECT_ID" "EIN"
## [3] "NAME" "TAXYR"
## [5] "FORMTYPE" "URL"
## [7] "EIN.1" "OrganizationTypeCd"
## [9] "GoverningDocumentListedInd" "SupportAmt"
## [11] "OtherSupportAmt" "BusinessNameLine1Txt"
## [13] "BusinessNameLine2Txt"
Check for xpaths that were used in the tables but do not exist in the Concordance file.
<- dplyr::filter( concordance, rdb_table == table.name )
c.table <- c.table$xpath
xpath.concordance setdiff( all.xpaths, xpath.concordance )
## [1] ""
## [2] "/Return/ReturnData/IRS990ScheduleA/SupportedOrgInformationGrp/EIN"
## [3] "/Return/ReturnData/IRS990ScheduleA/SupportedOrgInformationGrp/GoverningDocumentListedInd"
## [4] "/Return/ReturnData/IRS990ScheduleA/SupportedOrgInformationGrp/OrganizationTypeCd"
## [5] "/Return/ReturnData/IRS990ScheduleA/SupportedOrgInformationGrp/OtherSupportAmt"
## [6] "/Return/ReturnData/IRS990ScheduleA/SupportedOrgInformationGrp/SupportAmt"
## [7] "/Return/ReturnData/IRS990ScheduleA/SupportedOrgInformationGrp/SupportedOrganizationName"
## [8] "/Return/ReturnData/IRS990ScheduleA/SupportedOrgInformationGrp/SupportedOrganizationName/BusinessNameLine1Txt"
## [9] "/Return/ReturnData/IRS990ScheduleA/SupportedOrgInformationGrp/SupportedOrganizationName/BusinessNameLine2Txt"
Capture to update the Concordance file.
<- setdiff( all.xpaths, xpath.concordance )
these cat( these, sep="\n", file=paste0( folder.name, "/MISSING-XPATHS.txt" ) )
Check the status of failed URLs.
if( length(fail.log) > 0 )
{ sapply( fail.log, get_message ) %>%
table() %>%
pander()
else {
} print("ALL URLs ARE VALID")
}
## [1] "ALL URLs ARE VALID"
The create_edgelist() functions v1 and v2 are in R/utils.R.
V1 retains the root nodes Return and ReturnData.
V2 drops the root nodes and uses form versions “990” and “990EZ” as the root to visualize the sets of tables separately.
# KEEP ROOT NOTES Return & ReturnData
<- create_edgelist_v1( all.xpaths ) el
Edgelist format:
V1 | V2 |
---|---|
Return | ReturnData |
ReturnData | IRS990 |
IRS990 | Form990PartVIISectionAGrp |
Form990PartVIISectionAGrp | PersonNm |
Form990PartVIISectionAGrp | TitleTxt |
Form990PartVIISectionAGrp | AverageHoursPerWeekRt |
Form990PartVIISectionAGrp | IndividualTrusteeOrDirectorInd |
Form990PartVIISectionAGrp | OfficerInd |
Form990PartVIISectionAGrp | ReportableCompFromOrgAmt |
Form990PartVIISectionAGrp | ReportableCompFromRltdOrgAmt |
# library( data.tree )
<- FromDataFrameNetwork( network=el )
nd print( nd )
## levelName
## 1 Return
## 2 °--ReturnData
## 3 °--IRS990ScheduleA
## 4 °--SupportedOrgInformationGrp
## 5 ¦--EIN
## 6 ¦--GoverningDocumentListedInd
## 7 ¦--OrganizationTypeCd
## 8 ¦--OtherSupportAmt
## 9 ¦--SupportAmt
## 10 °--SupportedOrganizationName
## 11 ¦--BusinessNameLine1Txt
## 12 °--BusinessNameLine2Txt
SetGraphStyle( nd, rankdir = "LR")
SetEdgeStyle( nd, arrowhead = "vee", color = "grey20", penwidth = 2 )
SetNodeStyle( nd,
style = "filled,rounded",
shape = "box",
fillcolor = "LightBlue",
fontname = "helvetica",
fontcolor="black",
tooltip = GetDefaultTooltip )
# SetNodeStyle(acme$IT, fillcolor = "LightBlue", penwidth = "5px")
plot( nd )
# library( networkD3 )
# net <- ToDataFrameNetwork( nd, "name" )
# simpleNetwork( net, fontSize = 12 )
simpleNetwork( el, fontSize = 12 )
#plot with networkD3
<- ToListExplicit( nd, unname = TRUE )
radnet radialNetwork( radnet, fontSize = 12 )
<-
t.xpaths %>%
concordance filter( rdb_table == table.name ) %>%
select( xpath ) %>%
unique() %>%
arrange( xpath ) %>%
mutate( id=row_number() )
%>%
t.xpaths kable( align="l" )
xpath | id |
---|---|
/Return/ReturnData/IRS990ScheduleA/Form990ScheduleAPartI/SupportedOrgInformation/OrganizedInUS | 1 |
/Return/ReturnData/IRS990ScheduleA/Form990ScheduleAPartI/SupportedOrgInformation/SupportedOrganizationNotified | 2 |
/Return/ReturnData/IRS990ScheduleA/SupportedOrgInformation/OrganizedInUS | 3 |
/Return/ReturnData/IRS990ScheduleA/SupportedOrgInformation/SupportedOrganizationNotified | 4 |
/Return/ReturnData/IRS990ScheduleA/SupportedOrgInformationGrp/SupportedOrgNotifiedInd | 5 |
/Return/ReturnData/IRS990ScheduleA/SupportedOrgInformationGrp/USOrganizedInd | 6 |
Identify all unique groups used in xpaths.
if( is.null(table.headers) )
find_table_headers( table.name ) } else
{ table.headers
## [1] "//SupportedOrgInformationGrp/SupportedOrganizationName"
## [2] "//SupportedOrgInformationGrp/EIN"
## [3] "//SupportedOrgInformationGrp/OrganizationTypeCd"
## [4] "//SupportedOrgInformationGrp/OrganizationTypeDesc"
## [5] "//SupportedOrgInformationGrp/GoverningDocumentListedInd"
## [6] "//SupportedOrgInformationGrp/SupportAmt"
## [7] "//SupportedOrgInformationGrp/OtherSupportAmt"
## [8] "//SupportedOrgInformationGrp/SupportedOrgNotifiedInd"
## [9] "//SupportedOrgInformationGrp/USOrganizedInd"
## [10] "//Form990ScheduleAPartI/Church170b1Ai"
## [11] "//Form990ScheduleAPartI/School170b1Aii"
## [12] "//Form990ScheduleAPartI/Hospital170b1Aiii"
## [13] "//Form990ScheduleAPartI/HospitalNameAndAddress"
## [14] "//Form990ScheduleAPartI/MedicalResearchOrg170b1Aiii"
## [15] "//Form990ScheduleAPartI/CollegeSupportOrg170b1Aiv"
## [16] "//Form990ScheduleAPartI/GovernmentalUnit170b1Av"
## [17] "//Form990ScheduleAPartI/PubliclySupportedOrg170b1Avi"
## [18] "//Form990ScheduleAPartI/CommunityTrust170b1Avi"
## [19] "//Form990ScheduleAPartI/PubliclySupportedOrg509a2"
## [20] "//Form990ScheduleAPartI/TestingForPublicSafety509a4"
## [21] "//Form990ScheduleAPartI/SupportingOrganization509a3"
## [22] "//Form990ScheduleAPartI/SupportingOrg509a3Type1"
## [23] "//Form990ScheduleAPartI/SupportingOrg509a3Type2"
## [24] "//Form990ScheduleAPartI/SupportingOrg509a3Type3FuncInt"
## [25] "//Form990ScheduleAPartI/WrittenIRSDetermination"
## [26] "//Form990ScheduleAPartI/SupportedOrgInformation"
## [27] "//Form990ScheduleAPartI/SumOfAmounts"
## [28] "//Form990ScheduleAPartI/SupportingOrg509a3Type3Other"
## [29] "//Form990ScheduleAPartI/CertificationCheckbox"
## [30] "//Form990ScheduleAPartI/ContribByControllingIndividual"
## [31] "//Form990ScheduleAPartI/ContributionByRelatedParty"
## [32] "//Form990ScheduleAPartI/ContribBy35ControlledEntity"
## [33] "//SupportedOrgInformation/Form990ScheduleAPartI"
## [34] "//SupportedOrgInformation/SupportedOrgInformation"
Groups can then be used to extract the 1:M RDB tables:
if( is.null(table.headers) )
<- find_table_headers( table.name ) }
{ table.headers <- "https://nccs-efile.s3.us-east-1.amazonaws.com/xml/201913079349300206_public.xml"
demo.url <- xml2::read_xml( demo.url )
doc ::xml_ns_strip( doc )
xml2<- get_table_v2( doc, table.name, table.headers )
df.demo df.demo
## NULL
Mapping of xml variable names to concordance variable names:
get_var_map( table.name=table.name )
A table with proper names:
<- get_var_map( table.name=table.name )
v.map
if( ! is.null(df.demo) )
re_name( df.demo, v.map ) %>% as.data.frame()} {
Demo of functions - demo data only.
[[1]]
<Form990PartVIISectionAGrp>
<PersonNm>
{text}
<TitleTxt>
{text}
<AverageHoursPerWeekRt>
{text}
<IndividualTrusteeOrDirectorInd>
{text}
<OfficerInd>
{text}
<ReportableCompFromOrgAmt>
{text}
<ReportableCompFromRltdOrgAmt>
{text}
<OtherCompensationAmt>
{text}
1]] %>% xml_parent() %>% xmltools::xml_view_trees() node.list[[
(1) ------------
+-- Form990PartVIISectionAGrp
+-- PersonNm
+-- TitleTxt
+-- AverageHoursPerWeekRt
+-- IndividualTrusteeOrDirectorInd
+-- OfficerInd
+-- ReportableCompFromOrgAmt
+-- ReportableCompFromRltdOrgAmt
++-- OtherCompensationAmt
(2) ------------
+-- Form990PartVIISectionAGrp
+-- PersonNm
+-- TitleTxt
+-- AverageHoursPerWeekRt
+-- IndividualTrusteeOrDirectorInd
+-- OfficerInd
+-- ReportableCompFromOrgAmt
+-- ReportableCompFromRltdOrgAmt
++-- OtherCompensationAmt